{"exhibit_id": "DX-1017", "exhibit": "DX 1017", "party": "Defense", "type": "Text messages", "admitted_trial_day": "Day 8 (May 6, 2026)", "uploaded_box_pt": "2026-05-06T14:57:00", "uploader": "Morrison Foerster", "pages": 3, "size_bytes": 115664, "source_pdf": "DX-1017.pdf", "pdf_url": "https://media.mts-in.com/DX-1017.pdf", "body_markdown": "# DX 1017 — Feb 25, 2023 Zilis → friend \"Have to resign openai board btw / E's effort has become well known\"\n\n> Shivon Zilis's February 25, 2023 text exchange with a friend (handle \"Shahini Rubicon Fluffer\") on the day Sam Altman called her to tell her she had to leave the OpenAI board because **\"the father of your babies starts a competitive effort and will recruit out of openai there is nothing to be done\"** — a contemporaneous record of Zilis's resignation in the run-up to xAI's launch.\n\n## Document type\n**Text messages, Cellebrite-style \"Short Message Report\" extraction.** Three pages: cover sheet (1 conversation, 24 messages, 3 participants — \"Shahini Rubicon Fluffer\" and Shivon Zilis appearing twice; handles redacted); page 2 captures messages timestamped 2/25/2023 8:10–8:15 PM; page 3 is blank/fully redacted. Bates ZILIS-0000937–0000939. Document ID DX-1017.\n\n## Logistics\n- **Trial admission:** Day 8 (May 6, 2026) — used during Zilis live direct examination. No prior wiki reference; admission day inferred from Box upload date.\n- **Box upload:** 2026-05-06 14:57:00 PT — Day 8 mid-afternoon defense batch (clustered with [[DX 627]], [[DX 651]], [[DX 712]], [[DX 1290]]).\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~113 KB, 3 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `1017.pdf`.\n\n## Transcribed text\n\n> **Conversation:** dd47762fe7a2a5e3a20ab888a4ee436f\n> **Date Range:** 2/25/2023\n> **Participants:** \"Shahini Rubicon Fluffer\" [redacted] · Shivon Zilis [redacted] · Shivon Zilis [redacted]\n>\n> *(Page 2 visible messages, all 2/25/2023:)*\n>\n> **SZ** (Shivon Zilis) — 8:10 PM\n> Have to resign openai board btw\n>\n> **SZ** — 8:10 PM\n> E's effort has become well known\n>\n> **SF** (Shahini Rubicon Fluffer) — 8:11 PM\n> Fuck.\n>\n> **SF** — 8:11 PM\n> You ok?\n>\n> **SZ** — 8:11 PM\n> Yeah. Sam called this morning and I knew what it was about before he called\n>\n> **SZ** — 8:12 PM\n> When the father of your babies starts a competitive effort and will recruit out of openai there is nothing to be done\n>\n> **SF** — 8:12 PM\n> Understandable but I'm still bummed for you. E should put you on the board of the new thing…\n>\n> **SZ** — 8:14 PM\n> Yeah we had a talk and I did share that I'm bummed because it was a nice way to maintain contribution while raising kids\n>\n> **SZ** — 8:15 PM\n> He proactively apologized that he had pruned my friend network through this\n>\n> *(Page 3: messages redacted in production)*\n\n## Commentary\n\nDX 1017 is a tight, high-impact Day-8 Zilis exhibit. Three lines get airtime: (i) **\"E's effort has become well known\"** — by February 25, 2023, Musk's competing AI venture (the proto-xAI effort) was an open secret in OpenAI circles, **15 months before** xAI was formally announced (March 2023 / July 2023) and **a year before** Musk filed this lawsuit; (ii) **\"the father of your babies starts a competitive effort and will recruit out of openai there is nothing to be done\"** — the operative reason for Zilis's resignation, in her own words: it was a *competitive* conflict, not a mission conflict; and (iii) **\"He proactively apologized that he had pruned my friend network through this.\"** Defense uses these three lines to argue that the ostensibly philanthropic litigation Musk filed in 2024 was the **continuation** of a competitive AI venture he had been quietly building since at least early 2023 — the period plaintiffs frame as \"phase three\" disillusionment but defense frames as Musk-launches-rival-and-sues-incumbent. Cross-reference: [[DX 1156]] (Feb. 2025 Musk-Zuckerberg \"bidding on the OpenAI IP\" texts) and [[DX 1157]] (Feb. 10, 2025 Toberoff $97B bid letter).\n\n---\n*See also:* [[Day 8|Day 8 digest]] · [[Shivon Zilis]] · [[Sam Altman]] · [[DX 1156]] · [[DX 1157]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-1025", "exhibit": "DX 1025", "party": "Defendants", "type": "Tweet", "admitted_trial_day": "Day 3 (April 29, 2026)", "uploaded_box_pt": "2026-04-29T15:39:12", "uploader": "Morrison Foerster", "pages": 1, "size_bytes": 539648, "source_pdf": "DX-1025.pdf", "pdf_url": "https://media.mts-in.com/DX-1025.pdf", "body_markdown": "# DX 1025 — Mar 15, 2023 Musk tweet: \"I'm still confused as to how a non-profit to which I donated ~$100M somehow became a $30B market cap for-profit. If this is legal, why doesn't everyone do it?\"\n\n> A single Twitter / X post by @elonmusk dated 12:49 PM PT on March 15, 2023 — eight months before Musk would file the original Musk v. Altman complaint (Feb 29, 2024) — claiming he had donated approximately **$100 million** to OpenAI as a nonprofit. The screenshot shows 16 million views, 141K likes, 15K retweets, 6.5K replies. At deposition (per [[Day 3|Day 3 digest]]), Musk testified he was **\"mistaken\"** about the $100M figure; the Musk Foundation's actual cumulative giving to OpenAI was lower.\n\n## Document type\n**Tweet.** Single screenshot of one X post by verified account @elonmusk, capturing username, post text, timestamp, view/engagement counts, and the standard X post-action bar (reply, repost, like, bookmark, share). Bates OPENAI_MUSK00039675.\n\n## Logistics\n- **Trial admission:** Day 3 (April 29, 2026) — Musk direct concludes; cross begins. Per the chunk wiki note: \"**DX 1025 — Mar 2023 Musk $100M tweet (depo: I was mistaken).**\" The exhibit is the defense's primary tool for impeaching Musk's contemporaneous public statements about his donations to OpenAI: the actual figure has been variously stated as ~$45M (per the Musk Foundation [[PX 91]] Fidelity Charitable grant compendium) or ~$50M (per Musk's own deposition testimony). The \"$100M\" tweet — viewed 16 million times — is the headline number Musk's lawsuit ultimately rested on, and it was wrong by roughly half.\n- **Box upload:** 2026-04-29 15:39:12 PT — Day 3 mid-afternoon batch.\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~527 KB, 1 page.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `1025 .pdf`.\n\n## Transcribed text\n\n> **Elon Musk** ✓ ☒\n> **@elonmusk**\n>\n> I'm still confused as to how a non-profit to which I donated ~$100M somehow became a $30B market cap for-profit. If this is legal, why doesn't everyone do it?\n>\n> 12:49 PM · Mar 15, 2023 · **16M Views**\n>\n> ↩ 6.5K · ⟲ 15K · ♥ 141K · 🔖 1.8K · ↗\n\n## Commentary\n\nDX 1025 is a small piece of paper doing very heavy work for the defense. **Defense's read:** the **$100M** figure is the public claim Musk used to seed years of \"stole my charity\" rhetoric — culminating in the original Musk v. Altman complaint (Feb 29, 2024) and this very lawsuit. At his deposition (used on cross by [[OpenAI counsel]] / Schubert on Day 3), Musk **conceded he was \"mistaken\"** — the actual Musk-side giving to OpenAI nonprofit was substantially less than $100M, and a meaningful portion was routed through a Fidelity Charitable [[donor-advised fund]] rather than directly from Musk personally (see [[PX 91]] — the Fidelity Charitable grant letter compendium, totaling roughly $45M in DAF-routed grants 2017–2020). The \"$30B market cap\" figure was also speculative as of March 2023 — OpenAI's January 2023 tender at a $29B valuation was a secondary, not a primary, market signal. **Plaintiffs' read:** the precise dollar figure is immaterial to the legal claim; whether Musk donated $44M, $50M, or $100M, the structural transformation of a 501(c)(3) into a $30B-(now $300B+)-market-cap for-profit raises the same charitable-trust and unjust-enrichment questions. Musk's \"mistaken\" deposition concession is a narrow factual correction, not a withdrawal of the legal theory. The \"If this is legal, why doesn't everyone do it?\" line is, plaintiffs argue, the actual heart of the case — a public articulation of the very harm the **California Attorney General** has authority to police (see [[Day 9|Day 9 digest]] § standing). The 16M views and 141K likes establish the tweet's reach: this is the framing that shaped public and regulatory perception of OpenAI's conversion. The exhibit also tees up the defense's broader credibility narrative — that Musk's public statements about OpenAI have been imprecise, retrospective, and shaped by his post-2023 competitive interest in [[xAI]] (founded July 2023, four months after this tweet).\n\n---\n*See also:* [[Day 3|Day 3 digest]] · [[PX 91]] · [[PX 156]] · [[PX 157]] · [[Elon Musk]] · [[Sam Altman]] · [[xAI]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-1040", "exhibit": "DX 1040", "party": "Defense", "type": "Press release", "admitted_trial_day": "unknown", "pages": 2, "size_bytes": 502469, "source_pdf": "DX-1040.pdf", "bates": "OPENAI_MUSK00039061-39062", "pdf_url": "https://media.mts-in.com/DX-1040.pdf", "body_markdown": "# DX 1040 — OpenAI's November 17, 2023 firing announcement (\"OpenAI announces leadership transition\")\n\n> **The official OpenAI press release** announcing Sam Altman's removal as CEO, dated **November 17, 2023**. The line that ricocheted around the world: Altman \"**was not consistently candid in his communications with the board**.\" Mira Murati appointed interim CEO; Greg Brockman steps down as chairman (but remains at the company at that point).\n\n## Document type\n**Press release** — public-facing OpenAI announcement, headlined \"**OpenAI announces leadership transition**.\" Bates **OPENAI_MUSK00039061–39062**.\n\n## Logistics\n- **Trial admission:** unknown (DX-1040 not yet matched in the public record).\n- **File size:** ~500 KB, 2 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `1040.pdf`.\n\n## Transcribed text\n\n> **November 17, 2023 — Company**\n>\n> ## OpenAI announces leadership transition\n>\n> *Chief technology officer Mira Murati appointed interim CEO to lead OpenAI; Sam Altman departs the company.*\n>\n> *Search process underway to identify permanent successor.*\n>\n> The board of directors of OpenAI, Inc., the 501(c)(3) that acts as the overall governing body for all OpenAI activities, today announced that **Sam Altman will depart as CEO and leave the board of directors**. Mira Murati, the company's chief technology officer, will serve as interim CEO, effective immediately.\n>\n> A member of OpenAI's leadership team for five years, Mira has played a critical role in OpenAI's evolution into a global AI leader. She brings a unique skill set, understanding of the company's values, operations, and business, and already leads the company's research, product, and safety functions. Given her long tenure and close engagement with all aspects of the company, including her experience in AI governance and policy, the board believes she is uniquely qualified for the role and anticipates a seamless transition while it conducts a formal search for a permanent CEO.\n>\n> **Mr. Altman's departure follows a deliberative review process by the board, which concluded that he was not consistently candid in his communications with the board, hindering its ability to exercise its responsibilities. The board no longer has confidence in his ability to continue leading OpenAI.**\n>\n> In a statement, the board of directors said: \"**OpenAI was deliberately structured to advance our mission: to ensure that artificial general intelligence benefits all humanity. The board remains fully committed to serving this mission.** We are grateful for Sam's many contributions to the founding and growth of OpenAI. At the same time, we believe new leadership is necessary as we move forward. As the leader of the company's research, product, and safety functions, Mira is exceptionally qualified to step into the role of interim CEO. We have the utmost confidence in her ability to lead OpenAI during this transition period.\"\n>\n> OpenAI's board of directors consists of OpenAI chief scientist **Ilya Sutskever**, independent directors Quora CEO **Adam D'Angelo**, technology entrepreneur **Tasha McCauley**, and Georgetown Center for Security and Emerging Technology's **Helen Toner**.\n>\n> As a part of this transition, **Greg Brockman will be stepping down as chairman of the board and will remain in his role at the company, reporting to the CEO**.\n>\n> OpenAI was founded as a non-profit in 2015 with the core mission of ensuring that artificial general intelligence benefits all of humanity. **In 2019, OpenAI restructured to ensure that the company could raise capital in pursuit of this mission, while preserving the nonprofit's mission, governance, and oversight. The majority of the board is independent, and the independent directors do not hold equity in OpenAI.** While the company has experienced dramatic growth, it remains the fundamental governance responsibility of the board to advance OpenAI's mission and preserve the principles of its Charter.\n\n## Commentary\n\nDX 1040 is the **single most-quoted document in the November 2023 firing record**. The \"**not consistently candid in his communications with the board**\" line is the authoritative text — the language the press, the employee letter to the board ([[DX 1046]]), and every subsequent court filing have wrapped around. The defense marks it as a DX (rather than letting plaintiffs anchor it) for a reason: the second-to-last paragraph contains the **affirmative defense framework** — that the 2019 restructure \"**preserve[d] the nonprofit's mission, governance, and oversight**,\" and that \"**the independent directors do not hold equity in OpenAI**.\" That paragraph is the structural rebuttal to plaintiffs' \"stole the charity\" theme — captured in OpenAI's own contemporaneous voice on the day of the firing.\n\nNote also what's *not* here: the announcement does not name the firing as related to safety, product, or finances. Tasha McCauley confirmed that framing on Day 9: \"**We weren't concerned that any product had been unsafely released in particular**. The work we had concerns about [was] **processes not being respected**.\" (See [[Day 9|Day 9 digest]] §2c.) The press release is, by design, a candor-and-governance document — and the words on this page anchor every subsequent witness's testimony about why the board acted.\n\n---\n*See also:* [[Day 8]] · [[Day 9]] · [[DX 1046]] · [[DX 1048]] · [[PX 304]] · [[PX 306]] · [[Mira Murati]] · [[Helen Toner]] · [[Tasha McCauley]] · [[Sam Altman]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-1046", "exhibit": "DX 1046", "party": "Defense (OpenAI/MS)", "type": "Letter", "admitted_trial_day": "Day 8 (May 6, 2026)", "uploaded_box_pt": "2026-05-06T14:57:01", "uploader": "OpenAI startup-fund defense firm", "pages": 15, "size_bytes": 910320, "source_pdf": "DX-1046.pdf", "pdf_url": "https://media.mts-in.com/DX-1046.pdf", "body_markdown": "# DX 1046 — November 2023 OpenAI employee letter to the board (\"unless all current board members resign\")\n\n> The mass-signed November 2023 OpenAI employee letter demanding the board resign and reinstate Sam Altman and Greg Brockman, threatening to decamp to Microsoft's newly announced subsidiary — the contemporaneous record of the firing-reversal pressure campaign that runs through Murati's, Zilis's, and Toner's Day 8 testimony.\n\n## Document type\n**Letter (open / multi-signatory employee letter).** Typed letter addressed \"To the Board of Directors at OpenAI,\" followed by 751 numbered employee signatures. CONFIDENTIAL stamp; production Bates OPENAI_MUSK00037737–OPENAI_MUSK00037751.\n\n## Logistics\n- **Trial admission:** Day 8 (May 6, 2026) — Murati live; Zilis live; Toner video begins.\n- **Box upload:** 2026-05-06 14:57:01 PT — uploaded the same day it came in at trial.\n- **Uploader:** OpenAI startup-fund defense firm.\n- **File size:** ~889 KB, 15 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `1046.pdf`.\n\n## Transcribed text\n\n> To the Board of Directors at OpenAI,\n>\n> OpenAI is the world's leading AI company. We, the employees of OpenAI, have developed the best models and pushed the field to new frontiers. Our work on AI safety and governance shapes global norms. The products we built are used by millions of people around the world. Until now, the company we work for and cherish has never been in a stronger position.\n>\n> The process through which you terminated Sam Altman and removed Greg Brockman from the board has jeopardized all of this work and undermined our mission and company. Your conduct has made it clear you did not have the competence to oversee OpenAI.\n>\n> When we all unexpectedly learned of your decision, the leadership team of OpenAI acted swiftly to stabilize the company. They carefully listened to your concerns and tried to cooperate with you on all grounds. Despite many requests for specific facts for your allegations, you have never provided any written evidence. They also increasingly realized you were not capable of carrying out your duties, and were negotiating in bad faith.\n>\n> The leadership team suggested that the most stabilizing path forward — the one that would best serve our mission, company, stakeholders, employees and the public — would be for you to resign and put in place a qualified board that could lead the company forward in stability. Leadership worked with you around the clock to find a mutually agreeable outcome. Yet within two days of your initial decision, you again replaced interim CEO Mira Murati against the best interests of the company. You also informed the leadership team that allowing the company to be destroyed \"would be consistent with the mission.\"\n>\n> **Your actions have made it obvious that you are incapable of overseeing OpenAI. We are unable to work for or with people that lack competence, judgement and care for our mission and employees. We, the undersigned, may choose to resign from OpenAI and join the newly announced Microsoft subsidiary run by Sam Altman and Greg Brockman. Microsoft has assured us that there are positions for all OpenAI employees at this new subsidiary should we choose to join. We will take this step imminently, unless all current board members resign, and the board appoints two new lead independent directors, such as Bret Taylor and Will Hurd, and reinstates Sam Altman and Greg Brockman.**\n>\n> 1. Mira Murati\n> 2. Brad Lightcap\n> 3. Jason Kwon\n> 4. Wojciech Zaremba\n> 5. Alec Radford\n> 6. Anna Makanju\n> 7. Bob McGrew\n> 8. Srinivas Narayanan\n> 9. Che Chang\n> 10. Lillian Weng\n> 11. Mark Chen\n> 12. Ilya Sutskever\n> 13. John Schulman\n> 14. Barret Zoph\n>\n> [Signatures 15–751 follow alphabetically across pages 2–15: Nick Ryder, Alex Paino, Miles Brundage, Matt Knight, Peter Welinder, Hannah Wong, Diane Yoon, William Fedus, Arka Dhar, Peter X. Deng, Jonathan Lachman, Aaron Hurst, Abe Land, Abigail Raman, Abram Wiesman, Adam Goldberg, Adam Goucher, Adam Groth, Adam Hupp, Adam Nace, Adam Perelman, Aditya Ramesh, Adrian Spânu, Adrien Ecoffet, Afsha Shaikh, AJ Ostrow, Ahmed El-Kishky, Aidan Clark, Akila Welihinda, Akshay Nathan, Alan Hickey, Aleksandr Borzunov, Alex Baker-Whitcomb, Alex Beutel, Alex Carney, Alex Chow, Alex Gee, Alex Iftimie, Alex Nichol, Alex Renzin, Alexander Karpenko, Alexei Karpenko, Alexey Ivanov, Alexander King, Alexander Kirillov, Alexandre Tachard Passos, Alexis Conneau, Ali Kamali, Alice Lee, Aliisa Rosenthal, Alison Harmon, Alison McPhail, Allan Jabri, Allie Teague, Allison Tam, Alvin (Jiahua) Wang, Amanda Anderson, Amanda Tenanes, Amin Tootoonchian, Amol Shah, Ananya Kumar, Andre Saraiva, Andrea Vallone, Andrej Karpathy, Andrew Braunstein, Andrew Cann, Andrew Kondrich, Andrew Peng, Andrew Top, Andrew Tulloch, Andrew YuXuan Liu, Andrew Busker, Andrey Mishchenko, Andy Brown, Angel Demirev, Angela Jiang, Angie Luo, Anish Tondwalkar, Ankush Gupta, Anna-Luisa Brakman, Anna McKean, Anna Tifft, Anthony Malinao, Anton Tananaev, Anushree Agrawal, Artemis Seaford, Arun Vijayvergiya, Arushi Sardana, Arvind Neelakantan, Arzav Jain, Ashley Pantuliano, Ashvin Nair, Atty Eleti, Avi Nayak, Avital Oliver, Barak Michener, Rebecca Waite, Behrooz Ghorbani, Belinda Truong, Ben Leimberger, Ben Rossen, Ben Wang, Benjamin Burrell, Benjamin Sokolowsky, Benjamin Zweig, Bertrand Fan, Bianca Martin, Bill Peebles, Bill Stavroulakis, Billie Jonn, Blake Samic, Bob Rotsted, Bobby Wu, Bogo Giertler, Boris Power, Bowen Baker, Boyang Niu, Brad Biegler, Brandon Wang, Brandon Houghton, Brendan Herron, Brendan Quinn, Brian Kelleher, Brittany Carey, Brittany Smith, Britton Jamison, Brock Whittaker, Brooke Chan, Bryan Ashley, Brydon Eastman, Cadran Cowansage, Caitlin Carroll, Cameron Raymond, Camillo Lugaresi, Carl Ross, Cary Bassin, Casey Chu, Catherine Yeh, Chak Ming Li, Chan Park, Coley Czarnecki, Corey Nayden, Channing Conger, Chelsea Carlson, Chelsea Voss, Chen Ding, Chia Chung (Yoshi) Wang, Chong Zhang, Chris Berner, Chris Brown, Chris Clark, Chris Finlayson, Chris Haugli, Christian Gibson, Christian Sotto, Christina Kim, Christina Strong, Christine Choi, Christine Nguyen, Chris Koch, Christine McLeavey, Chris Hallacy, Christopher Hesse, Cindy Yong, CJ Minott, CJ Weinmann, Clarence Ng, Clemens Winter, Colin Jarvis, Colin Seale, Cory Decareaux, Colin Wei, Conor Dragomanovich, Dalton Hubble, Damien Deville, Dan Fliflet, Dan Mossing, Daniel Adamson, Daniel Galin, Daniel Kappler, Daniel Levine, Daniel Levy, Daniel Selsam, Dante' Robinson, Daryl Neubieser, Dave Cummings, David Carr, David Clyde, David Dohan, David Duxin, David Farhi, David Hao, David Hengky, David Katz, David Kjelkerud, David Medina, David Mely, David Newman, David Robinson, David Sasaki, Davis Robertson, David Schnurr, David Smock, Davin Bogan, Denny Jin, Derek Chen, Derrick Tyler, Devon Lloyd, Diane Techer, Diogo Almeida, Dimitris Tsipras, Don Lee, Dominic Grillo, Dominick Lim, Douglas Li, Duc Phong Nguyen, Duncan Findlay, Dylan Royan Almeida, Edmund Wong, Edward J. Hu, Elena Chatziathanasiadou, Elie Georges, Eliot Hijano, Elizabeth Tseng, Elizabeth Olin, Elizabeth Proehl, Elizabeth Yang, Emma Lalley, Emma Redmond, Emma Tang, Emily Stern, Enoch Cheung, Erik Ritter, Eric Antonow, Eric Horacek, Eric Jiang, Eric Rynerson, Eric Sigler, Erica Lee, Erica Tom, Eric Luhman, Ehsan Asdar, Ethan Stock, Evan Lee, Evan Morikawa, Evan Weiss, Faiz Munshi, Farouk El Hamzawi, Fan Wang, Farzad Khorasani, Felipe Such, Felipe Torres, Fereshte Khani, Filipe de Avila Belbute Peres, Filippo Raso, Florencia Leoni Aleman, Foivos Tsimpourlas, Fotios Chantzis, Francis Real, Francis Song, Francis Zhang, Fred von Lohmann, Gabriel Goh, Garrett McCarthy, Gavin Ahern, Gene Oden, Geoffrey Salmon, Giambattista Parascandolo, Gideon Myles, Gildas Chabot, Glory Jain, Gretchen Krueger, Hadi Salman, Haidee Schwartz, Haiming Bao, Haitang Hu, Haozhun Jin, Harold Nguyen, Hart Andrin, Harri Edwards, Heather Schmidt, Heewoo Jun, Henk Tillman, Henri Roussez, Henrique Ponde de Oliveira Pinto, Henry Head, Hessam Bagherinezhad, Hongyu Ren, Huiwen Chang, Hugh Drinkwater, Hyung Won Chung, Hyunil (Chester) Cho, Hunter Lightman, Hyeonwoo Noh, Ian Kivlichan, Ian Mobbs, Ian O'Connell, Ian Osband, Ian Silber, Ian Sohl, Ibrahim Okuyucu, Ikai Lan, Ilan Bigio, Ilge Akkaya, Ilya Kostrikov, Ingmar Kanitscheide, Irwan Bello, Isa Fulford, Isabella D'Agostino, Ishaan Gulrajani, Jackie Hehir, (Jan) Hendrik Kirchner, Jacob Coxon, Jacob Menick, Jake Brill, Jake McNeil, Jamel Cooney, James Betker, Jamie Caulkins, James Dyett, James Hills, Jamie Kiros, Janine Korovesis, Jarad Gilmore, James Hairston, Janice Kwong, Jason Teplitz, Jason Tunzini, Jason Wei, Jason Wolfe, Javier Soto, Jean Harb, Jean Neaylon, Jeff Belgum, Jeff Chan, Jeff Harris, Jelani Robinson, Jenia Varavva, Jennifer Lau, Jeorge Corcoran, Jeremiah Currier, Jerry Tworek, Jessica James, Jessica Lee, Jessica Shieh, Jiacheng Feng, Jiahui Yu, Jiayi Weng, Jie Tang, Jieqi Yu, Jigar Bhati, Jillian Khoo, Joanne Jang, Jodi Kean, Joe Gillespie, Joe Palermo, Joe Taylor, Joel Parish, Joey Flynn, Johannes Heidecke, John Allard, John Lam, John Rizzo, Jonathan Ward, Jonathan Becton, Jonathan Gordon, Jonathan McKay, Jonathan Reiber, Jong Wook Kim, Joost Huizinga, Jordan Sitkin, Jordan Williams, Jos Kraaijeveld, Joscelyne Jauregui, Joseph Landers, Joseph Millman, Josh Gross, Josh Kaplan, Josh Snyder, Joshua Achiam, Joy Cashman, Joyce Lee, Juan \"Red\" Avila, Juntang Zhuang, Justin Boyle, Justin Lebar, Justin Wang, Juston Forte, Kai Chen, Kai Fricke, Kai Hayashi, Kai Xiao, Kaitlin Gottfried, Karen Toro, Karen Walsh, Karolis Kosas, Karl Cobbe, Karthik Rangarajan, Katarina Slama, Kathy Son, Katie Mayer, Katy Elkin, Kayla Wood, Kendra Rimbach, Ken Langlinais, Kenny Hsu, Kevin Button, Kevin Cavanagh, Kevin Gladstone, Kevin Liu, Kevin Metti, Kevin Park, Kevin Yu, Kia Hoxie, Kiel Howe, Kim Malfacini, Kleanthes Koniaris, Krissi Robinson, Kristen Dallara, Krithika Muthukumar, Kyla Sheppard, Kyle Miller, Kyle Luther, Lane Dilg, Lama Ahmad, Larry Kai, Larry Xu, Laura Wang, Lauren Itow, Lauren Workman, Lee Grimes, Lee Wei An, Leigh Ann Montano, Lenny Bogdonoff, Li Jing, Lia Guy, Lien Mamitsuka, Lin Yang, Lindsey held, Lisa Pan, Logan Kilpatrick, Louis Feuvrier, Loren Kwan, Lorenz Kuhn, Lu Zhang, Lucy Chen, Lukasz Kaiser, Lukasz Kondraciuk, Luke Metz, Luke Miller, Mada Aflak, Maddie Casey, Maddie Simens, Madelaine Boyd, Madeleine Thompson, Madeline Casey, Maggie Hott, Manas Joglekar, Marat Dukhan, Marc Manara, Marco Noce, Mario Saltarelli, Marko Tintor, Mark Hudnall, Mark Jaroszewski, Marvin Zhang, Mason Meyer, Mateusz Litwin, Matt Benice, Matt Beutler, Matt Castle, Matt Chung, Matt Hartill, Matt Yoshida, Matthew Lim, Max Reid, Meghan Shah, Mehmet Yatbaz, Mengyuan Yan, Michael Flossman, Michael Janner, Michael Kolhede, Michael McGrew, Michael Lampe, Michael Petrov, Mikhail Pavlov, Michael Schade, Michael Trinh, Michael Wu, Michelle Fradin, Michelle Kim, Michelle Pokrass, Michelle Zheng, Miguel Castro, Miguel Manriquez, Mike Heaton, Mike Hunter, Mike Pahre, Ming Chen, Mohammad Bavarian, Mollie Javerbaum, Molly Lin, Morgan Grafstein, Morgan Griffiths, Mostafa Rohaninejad, Naga Sai Anup Krishnamraju, Natalia Gimelshein, Natalie Summers, Natalie Staudacher, Natalie Cone, Natalie Kim, Neal Chuenprateep, Nick Cooper, Nick Hamilton, Nick Kartashov, Nick Turley, Nicholas Martin, Nick Lee, Nickolas Stathas, Nicolai Skabo, Nik Tezak, Nikki Sommer, Niko Felix, Nikunj Handa, Nitish Shirish Keskar, Noah Deutsch, Noah Jorgensen, Noam Brown, Noel Bundick, Ola Okelola, Oleg Boiko, Oleg Murk, Olivia Cheung, Olivier Godement, Orla McHenry, Owen Campbell-Moore, Owen Lin, Pamela Mishkin, Patricia Lue, Patryk Lesiewicz, Parijat Talkad, Patrik Goethe, Patrick Geonetta, Paul Baltescu, Paul McMillan, Paul Punia, Paul Shen, Pavel Izmailov, Pedram Keyani, Peter Dolan, Peter Hoeschele, Peter Zhokhov, Pei-Chee Juang, Phil Dibowitz, Philip Pronin, Philip Su, Philippe Tillet, Pia Santos, Phong Nguyen, Prafulla Dhariwal, Prasad Chakka, Pritesh Badiani, Qiming Yuan, Rachel Lee, Rachel Yun Shi Lim, Radhika Mathur, Raf Jakubanis, Rahul Arora, Rajan Troll, Rajeev Nayak, Randall Lin, Rapha Gontijo Lopes, Rasmus Rygaard, Raul Puri, Reiichiro Nakano, Reimar Leike, Reggie Hall, Renaud Gaubert, Renny Hwang, Reza Zamani, Rhythm Garg, Richard Hasslacher, Richard Ho, Richard Ngo, RJ Marsan, Rob Honsby, Robert Infantino, Robert Pulford, Rocky Smith, Rod Solaimani, Rohan Sahai, Rohit Ramchandani, Romain Huet, Roshan James, Rosie Campbell, Rory Carmichael, Rowan Zellers, Roy Ziv, Ruby Chen, Rui Shu, Ruslan Nigmatullin, Ryan Biddy, Ryan Cheu, Ryan Greene, Ryan Hartnett, Ryan Lowe, Ryan Peterson, Ryan Ragona, Sabrina Dermody, Sam Schoenholz, Sam Toizer, Samuel Wolrich, Sandhini Agarwal, Sarah Burns, Sandro Gianella, Sarah Shoker, Sarah Wray, Sarah Yoo, Saravpreet Hira, Scott Gray, Scott Lau, Scott Lessans, Scott Mayer McKinney, Sean Grove, Sean Metzger, Sean Murphy, Shantanu Jain, Sherry Lachman, Shyamal Anadkat, Shaun Van Weelden, Sheila Dunning, Shengjia Zhao, Shengli Hu, Sherwin Wu, Shibani Santurkar, Shino Jomoto, Shirong Wu, Shuyuan Zhang, Sierra Jackson, Simon Posada Fishman, Siyuan Fu, Soo Jin Park, Sophie Rose, Sowmya Ranganathan, Spencer Papay, Sriya Santhanam, Stacie Faggioli, Stefanie Biaggi, Stephanie Tran, Stephanie Lin, Stephen Petersilge, Steven Adler, Steven Bills, Steven Heidel, Stewart Hall, Suchir Balaji, Sully Chen, Tabarak Khan, Tal Broda, Tal Stramer, Tao Wang, Tao Xu, Tatiana Zolotova, Tarun Gogineni, Taylor Gordon, Teddy Lee, Tejal Patwardhan, Theresa Lopez, Thomas Degry, Thomas Dimson, Thomas Raoux, Tianhao Zheng, Tifa Chen, Tiffany Citra, Tim Brooks, Todd Underwood, Todor Markov, Toki Sherbakov, Tom Dupre la Tour, Tom Rubin, Tom Stasi, Tomer Kaftan, Tong Mu, Tony Wu, Trapit Bansal, Trevor Cai, Trevor Hildenberger, Tristan Heywood, Trevor Mavrakis, Troy Luhman, Troy Peterson, Tyna Eloundou, Ugurcan Turkdogan, Valerie Balcom, Vanessa Gatihi, Victoria Lampton, Victoria Spiegel, Vineet Kosaraju, Vinnie Monaco, Vishal Kuo, Vitchyr Pong, Wade Morgan, Warren Ouyang, Weiland Fong, Weiyi Zheng, Wenda Zhou, Wenlei Xie, Wesam Manassra, William Butler, Will DePue, Will Saborio, Xin Hu, Xinyang Geng, Xuanhan (Eimi) Zhang, Yang Song, Yash Patil, Yasu Sakamoto, Yilei Qian, Yining Chen, Yonadav Shavit, Yongjik Kim, Youlong Cheng, Yu Zhang, Yuchen He, Yuchen Zhang, Yufei Guo, Yujia Jin, Yunxing Dai, Jp Villafuerte, Yunxin Joy Jiao, Yury Malkov, Yuri Burda, Yutian Liu, Zahi Moudallal, Zhi Bie, Zheng Shao, Zhengwen Zhou.]\n\n## Commentary\n\nDX 1046 is the contemporaneous record of the November 2023 firing-reversal pressure campaign — the document Toner described from the witness stand as the moment Microsoft \"**was offering to hire away the entire OpenAI team, cushy jobs for anyone who wanted them**\" (`5/6/2026 Testimony @ ~15:02 PT`). Defense uses the letter to humanize the post-firing chaos and to anchor Murati's #1-signatory position; plaintiffs use the **same** letter as their cleanest evidence that the for-profit/Microsoft architecture has practical power to override the nonprofit board's mission-protection function — see [[Key Themes]] \"**It created a threat**.\" The bolded ultimatum paragraph is the on-the-page proof of plaintiffs' theory that OpenAI's nonprofit board, in 2023, had no functional levers left: 751 of its own employees told it to resign or watch the company decamp to Microsoft. That same dynamic returns through Toner's video on the **Deployment Safety Board** (\"alchemy, not chemistry\") and through Murati's \"**not always [candid]**\" answer about Altman. Cross-link to [[Quotes]] for the live readbacks.\n\n---\n*See also:* [[Day 8|Day 8 digest]] · [[Mira Murati]] · [[Helen Toner]] · [[Shivon Zilis]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-1048", "exhibit": "DX 1048", "party": "Defense", "type": "Text messages", "admitted_trial_day": "Day 8 (May 6, 2026)", "uploaded_box_pt": "2026-05-06T14:57:03", "uploader": "Morrison Foerster", "pages": 5, "size_bytes": 193207, "source_pdf": "DX-1048.pdf", "pdf_url": "https://media.mts-in.com/DX-1048.pdf", "body_markdown": "# DX 1048 — Mira Murati ↔ Satya Nadella iMessage thread, Nov 17–21, 2023 (defense-marked twin of PX 306)\n\n> The same November 17–21, 2023 Murati ↔ Nadella text thread that plaintiffs introduced as **PX 306** — re-marked here as a defense exhibit (DX 1048). Five pages of iMessages spanning the four days of the Sam Altman firing-and-return weekend, including Murati's \"**I'm not putting my name on this,**\" Nadella's \"**Thx. So sorry. But that is right,**\" and Murati's request for a public Microsoft statement so OpenAI would \"**not lose researchers to Demis or Elon**.\"\n\n## Document type\n**Text messages** — Apple iMessage screenshots between Mira Murati and Satya Nadella, exported as a 5-page PDF with timestamps. Bates MSFT_MUSK000065692. Marked HIGHLY CONFIDENTIAL. The same physical document is **dual-marked** as plaintiffs' PX 306 (uploaded Day 8 14:07:57 PT) and defense's DX 1048 (uploaded Day 8 14:57:03 PT).\n\n## Logistics\n- **Trial admission:** Day 8 (May 6, 2026) — Murati live testimony. Defense re-marks the document so it can be put to Murati for purposes the plaintiffs' marking would not cover (e.g., to highlight Murati's reliance on Microsoft as a unifying force, the **Microsoft-coordination** thread, and the contemporaneous absence of any complaint by Musk).\n- **Box upload:** 2026-05-06 14:57:03 PT — Day 8 mid-afternoon Defense batch (uploaded ~50 minutes after [[PX 306]]).\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~189 KB, 5 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `1048.pdf`.\n\n## Transcribed text\n\nSee [[PX 306]] for the full verbatim transcription of the Murati ↔ Nadella thread. Key passages defense highlights at the **DX 1048** marking:\n\n> **9/19/2017 — N/A** *(no, this is Nov 2023)*\n>\n> **Nov 17, 2023 — 3:33 PM Nadella:** \"Hi Mira… call when you have a chance. Thx Satya\"\n>\n> **Nov 19, 2023 — 7:52 PM Murati:** \"**I'm not putting my name on this**\"\n>\n> **Nov 20, 2023 — 1:20 AM Murati:** \"The petition we're sending says the following per your conversation with Sam and we wanted to confirm. '**Microsoft has assured us that there are positions for all OpenAI employees with the same compensation at this new subsidiary should we choose to join.**'\"\n>\n> **Nov 20, 2023 — 1:20 AM Murati:** \"We're removing it from public doc and communicating it privately\"\n>\n> **Nov 20, 2023 — 7:10 AM Nadella:** \"**Thx. So sorry. But that is right.**\"\n>\n> **Nov 20, 2023 — 10:44 AM Murati:** \"Satya could you please make a public statement soon that shows support for the joint openai team, basically bringing the team together? It's very important that we **don't lose researchers to Demis or Elon**. The technical team is being dragged in so many recruiting directions and a unified front would help immensely\"\n>\n> [Pages 2–5 continue the thread through Nov 21, 2023; full text available in [[PX 306]].]\n\n## Commentary\n\nDX 1048 is the **defense twin** of [[PX 306]] — same document, two markings, two trial uses. Defense uses the thread to (i) anchor Murati's testimony that \"**Microsoft and Satya in particular were important in unifying the team**\" (5/6/2026 Testimony @ ~10:13 PT) — i.e., Microsoft as a **stabilizing**, mission-protective force during the firing, which is the inverse of plaintiffs' \"captured by Microsoft\" framing; (ii) put on the record that the November 2023 reset was **driven by OpenAI's own staff petition**, not by Microsoft pulling the strings — Murati's \"we're sending\" / \"we wanted to confirm\" framing places agency with the OpenAI team; and (iii) highlight that Musk is referenced only as a **competing recruiter** (\"don't lose researchers to Demis or Elon\"), not as a charitable-trust beneficiary or an entity OpenAI was answering to. Plaintiffs' use of the same document is the inverse: Microsoft's confirmed willingness to absorb the entire OpenAI technical team is the precise leverage that Toner described on Day 8 as \"**a threat**\" (see [[Key Themes]] §\"Microsoft's role in Altman's reinstatement\"). Both readings are on the page.\n\n---\n*See also:* [[PX 306]] · [[Day 8|Day 8 digest]] · [[Mira Murati]] · [[Helen Toner]] · [[Sam Altman]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-1049", "exhibit": "DX 1049", "party": "Defense", "type": "Slack messages", "admitted_trial_day": "unknown", "pages": 2, "size_bytes": 2113811, "source_pdf": "DX-1049.pdf", "bates": "SUTSKEVER_MUSKSUB_00000525-526", "also_marked": "PX 319", "pdf_url": "https://media.mts-in.com/DX-1049.pdf", "body_markdown": "# DX 1049 — Helen Toner @channel Slack message announcing Emmett Shear as interim CEO (Nov 20, 2023, 1:42 AM UTC)\n\n> **The board's all-hands Slack message** sent at **1:42 AM UTC on November 20, 2023** (i.e., late Sunday night Pacific) — **three days after the firing**, after the weekend back-and-forth on Altman's possible reinstatement, announcing **Emmett Shear** as interim CEO and stating flatly: \"**Sam will not return as CEO.**\" Cross-marked as both **DX 1049** (defense) and **PX 319** (plaintiffs).\n\n## Document type\n**Slack message** (text, exported as a screen-capture style PDF). From: **Helen Toner**. Sent: **Mon 11/20/2023 1:42:00 AM (UTC)** — i.e., Sunday Nov 19, ~5:42 PM Pacific. To: **Ilya Sutskever**. Cc: **Adam D'Angelo**, **Tasha McCauley**. Subject: \"Slack message\". Marked **HIGHLY CONFIDENTIAL — ATTORNEYS' EYES ONLY**. Bates **SUTSKEVER_MUSKSUB_00000525–526**.\n\n## Logistics\n- **Trial admission:** unknown.\n- **File size:** ~2.1 MB, 2 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `1049.pdf`. Cross-marked under `319.pdf` as **PX 319**.\n\n## Transcribed text\n\n> **Conversation:** OpenAI #general (board @channel announcement)\n> **Date Range:** Mon 11/20/2023 1:42:00 AM (UTC)\n> **Participants:** Helen Toner → Ilya Sutskever (cc Adam D'Angelo, Tasha McCauley)\n>\n> **Helen Toner** — Mon 11/20/2023 1:42:00 AM (UTC)\n>\n> @channel, I want to share a message from the whole board to the team.\n>\n> We know this has been a difficult weekend and that it's been frustrating and confusing to figure out what is happening at OpenAI by reading about it on Twitter and in the media. We want to take a moment to update you on how we got to this point and where we are going from here.\n>\n> ### Our decision\n>\n> The board firmly stands by its decision as the only path to advance and defend the mission of OpenAI.\n>\n> Putting simply, **Sam's behavior and lack of transparency in his interactions with the board undermined the board's ability to effectively supervise the company in the manner it was mandated to do**. His departure was necessary to preserve the board's ability to execute its responsibilities and advance the mission of this organization.\n>\n> It is paramount that any CEO be honest and transparent with his or her board. We know that removing someone with Sam's high profile in our industry comes as a shock, but he lost the trust of the board of directors.\n>\n> We want to be abundantly clear: **We believe in OpenAI, its people, and its products.** This decision is not about product safety or security, the pace of development or OpenAI's finances. This was not about any singular incident. Bottom line, this was a governance issue that lies at the heart of how the board of this uniquely structured organization executes its responsibilities and advances its mission. **Despite rumors to the contrary, Sam will not return as CEO.**\n>\n> ### Why did the board act this way?\n>\n> OpenAI was deliberately structured as a non-profit entity governing the capped-profit entity. The board was purposefully established with the intention of independent oversight to advance OpenAI's mission, and we take our oversight responsibilities incredibly seriously. **Sam himself invited each board member into this role.** He has also explicitly acknowledged the importance of the board's independence and has publicly confirmed that the board has the authority to remove the CEO, if necessary.\n>\n> This decision was not made lightly. However, once we reached the decision, we felt the only way we could exercise our appropriate oversight and governance role was to take swift and decisive action. We apologize for the abruptness of the process that we felt was required by the situation. Even understanding the questions it has raised, we continue to believe our actions were necessary.\n>\n> ### What comes next?\n>\n> We have not spoken publicly since Friday, but we have been very active in private. We have been meeting with employees, investors and other stakeholders to ensure the best path forward for this incredible company. **We know that Sam's contributions to OpenAI went beyond technical leadership.** He had vision and the ability to activate the tech and investor community in support of our mission. And for that we are thankful.\n>\n> As you all know, we previously asked **Mira Murati** to serve as interim CEO while we searched for a long-term replacement. Given the developments over the past 48 hours, we have adjusted course. **We could not be more excited that Emmett Shear has agreed to join OpenAI as interim CEO.** Emmett is a well-established and trusted leader, having built and led Twitch from the ground-up over the past 16 years. He believes deeply in OpenAI's mission and has the unique mix of skills, expertise and relationships that will drive OpenAI forward.\n>\n> In our conversations in recent days, Emmett's name was repeatedly identified by leaders across the industry as the clear choice to lead OpenAI going forward.\n>\n> We want to thank Mira for agreeing to take on the interim CEO role, and we are incredibly pleased that she has agreed to stay on at the company in her role as Chief Technology Officer.\n>\n> We also want to thank **Satya Nadella and Microsoft**, who expressed their strong support for Emmett when we spoke with them earlier this evening.\n>\n> Emmett will have more to say about his vision for leading OpenAI shortly. We expect him to begin meeting with employees this evening, and to say more publicly in the coming days.\n>\n> The board has full confidence in Emmett, the incredible team at OpenAI and the company's ability to continue to succeed under new leadership. We look forward to supporting him and you through the next phase of OpenAI's growth.\n>\n> **Adam, Helen, Ilya and Tasha**\n> **OpenAI Board of Directors**\n\n## Commentary\n\nDX 1049 is the **board's defense of the firing in their own words**, written ~72 hours after [[DX 1040|the original Nov 17 announcement]] and at the *peak* of the reinstatement crisis — after the Microsoft \"open job offer\" overture and before the Tuesday agreement to bring Altman back. Three things make it load-bearing:\n\n1. **It locks in why the board acted.** \"**Sam's behavior and lack of transparency in his interactions with the board undermined the board's ability to effectively supervise the company.**\" Plaintiffs and defense both cite this — plaintiffs to anchor the candor problem, defense to lock in that the firing was about *governance*, \"**not about product safety or security, the pace of development or OpenAI's finances**\" and \"**not about any singular incident**.\" That language matches almost word-for-word what Tasha McCauley confirmed on Day 9 (see [[Day 9|Day 9 digest]] §2c).\n2. **It names Sutskever as the message's recipient.** The cc-list is **Adam D'Angelo, Tasha McCauley** — but the *to* line is **Ilya Sutskever**. By the time of this post, Sutskever was being sent a draft of the message *signed in his name*, even though the trial record reflects his eventual reversal of position the next day. The Bates source — `SUTSKEVER_MUSKSUB` — confirms the document came from Sutskever's production.\n3. **\"Sam will not return as CEO\"** + **Emmett Shear announcement.** This is the document where the board doubled down on the firing 72 hours in. By Tuesday, that position would invert — Altman returns, Toner/McCauley/D'Angelo resign, Shear's interim CEO tenure ends almost as soon as it began.\n\nThat this document is **cross-marked as PX 319** is itself the story: both sides put weight on it. Plaintiffs lean on the candor framework and the explicit framing of the for-profit as \"**capped-profit**\" governed by the nonprofit; defense leans on the \"**this is not about product safety or security**\" disclaimer and the description of the structure as \"**deliberately**\" designed.\n\n---\n*See also:* [[Day 8]] · [[Day 9]] · [[DX 1040]] · [[DX 1046]] · [[DX 1048]] · [[PX 304]] · [[PX 306]] · [[PX 319]] · [[Helen Toner]] · [[Tasha McCauley]] · [[Adam D'Angelo]] · [[Ilya Sutskever]] · [[Sam Altman]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-1083", "exhibit": "DX 1083", "party": "Defense", "type": "Text messages", "admitted_trial_day": "Day 4 (April 30, 2026)", "uploaded_box_pt": "2026-04-30T15:13:15", "uploader": "Morrison Foerster", "pages": 2, "size_bytes": 96014, "source_pdf": "DX-1083.pdf", "pdf_url": "https://media.mts-in.com/DX-1083.pdf", "body_markdown": "# DX 1083 — Dec 13, 2024 Zuckerberg → Musk, \"Meta sent a letter to the California AG supporting your lawsuit\"\n\n> A two-page, three-message text exchange between Mark Zuckerberg and Elon Musk on December 13, 2024 — Zuckerberg giving Musk a heads-up that \"**Meta sent a letter to the California AG supporting your lawsuit against OpenAI**\" before the leaked letter went public.\n\n## Document type\n**Text messages, Cellebrite-style \"Short Message Report\" extraction.** Two pages: cover sheet (1 conversation, 3 messages, 3 participants — two Elon Musk handles plus Mark Zuckerberg, all redacted) and one page of message-by-message transcript with timestamps in GMT-08:00. Bates 2024MUSK-0006235–6236. Marked Highly Confidential-AEO.\n\n## Logistics\n- **Trial admission:** Day 4 (April 30, 2026 — Musk cross + redirect; Birchall direct + cross). Per [[Key Themes|Day 4 quotes table]]: \"DX 1083 — Zuckerberg-Musk text: 'Quick heads up that Meta sent a letter to the California AG supporting your lawsuit'\" `(043026TT.txt:1626–1635)`.\n- **Box upload:** 2026-04-30 15:13:15 PT — Day 4 mid-afternoon defense batch (clustered with DX 539, DX 849, DX 1156, DX 1285, DX 516).\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~94 KB, 2 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `1083 .pdf` (note trailing space in filename).\n\n## Transcribed text\n\n> **Date Range:** 12/13/2024 — **Total Messages:** 3 — **Participants:** 3 (two Elon Musk handles, Mark Zuckerberg)\n> [Times in GMT -08:00]\n>\n> **Mark Zuckerberg** — 12/13/2024, 4:09 PM\n> Quick heads up that Meta sent a letter to the California AG supporting your lawsuit against OpenAI. Someone (not us) leaked leaked the letter and it will be public in the next hour. Wanted to make sure you heard this from me.\n>\n> *Receipts: Elon Musk read 12/13/2024, 4:19 PM*\n>\n> **Elon Musk** — 12/13/2024, 4:20 PM\n> Ok\n>\n> *Receipts: Mark Zuckerberg read 12/13/2024, 4:20 PM*\n>\n> **Mark Zuckerberg** — 12/13/2024, 6:24 PM\n> I have an idea to run by you. Not urgent, but let me know if there's a good time to call in the next few days.\n>\n> *Receipts: Elon Musk read 12/13/2024, 8:00 PM*\n\n## Commentary\n\nDX 1083 is the **prequel** to the [[DX-1156|February 2025 \"Are you open to the idea of bidding on the OpenAI IP with me and some others?\"]] text exchange. In December 2024 Zuckerberg gave Musk a friendly heads-up about Meta's letter to the California AG supporting Musk's lawsuit; six weeks later, Zuckerberg was on the [[Key Themes|$97.4B bid for OpenAI's IP]] syndicate. Defense uses DX 1083 for the same theme as DX 1156: Musk's stated charitable-trust theory is in tension with the **competing-AI-CEO support** (Meta) for the lawsuit, which on the defense's framing locates Musk's motive in **competitive harm to OpenAI** rather than mission protection. The \"**I have an idea to run by you**\" hanging hook in Zuckerberg's second message — six weeks before the IP bid — is the textual seed of the joint-bid plan. **Note:** the Zuckerberg \"(not us)\" leak disclaimer is a defamation hedge — Meta wanted Musk to know they did not leak the letter, only that the letter was about to become public. This dovetails with [[Mark Zuckerberg]]'s wiki page, which collects the DX 1083 / DX 1156 pair.\n\n---\n*See also:* [[Day 4|Day 4 digest]] · [[Mark Zuckerberg]] · [[DX-1156]] · [[Sam Altman]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-1092", "exhibit": "DX 1092", "party": "Defense", "type": "Email thread", "admitted_trial_day": "Off-day catch-up upload (uploaded 2026-05-02)", "uploaded_box_pt": "2026-05-02T09:41:20", "uploader": "Morrison Foerster", "pages": 2, "size_bytes": 455477, "source_pdf": "DX-1092.pdf", "pdf_url": "https://media.mts-in.com/DX-1092.pdf", "body_markdown": "# DX 1092 — Nov 2017 Pioneer Building security & access controls quotes (Birchall ↔ Clark ↔ Zilis)\n\n> Three-message thread (Nov 22 – Nov 29, 2017) between Chris Clark (OpenAI), Jared Birchall (Musk Foundation/Excession), and Shivon Zilis on **Pioneer Building security guard quotes** (Bannerman, Security Industry Specialists, Inter-State) and **building entry/parking-gate access controls**, with Clark's suggested payment mechanism: \"**bump up the monthly building lease grant Jared already has in place**.\"\n\n## Document type\n**Email thread, 2 pages, four messages.** Top message from Jared Birchall <jared@excession.com> to Chris Clark <chris@openai.com>, cc Shivon Zilis <shivon@neuralink.com>, sent Wed 29 Nov 2017 5:52:55 PM, subject \"Re: Quotes?\" Embeds Clark's earlier responses and the original Nov 22, 2017 forwarded \"Re: Quotes?\" message between Clark and Zilis with the security vendor matrix. Production-stamped Confidential; Bates EXMF-0000261–0000262. Document ID DX-1092.\n\n## Logistics\n- **Trial admission:** Uploaded 2026-05-02 09:41:20 PT — \"Off-day catch-up upload\" (the Saturday between Day 5 (May 1) and Day 6 (May 4)). Substantively pertinent to the [[Key Themes#The Pioneer Building|Pioneer Building]] thread of plaintiffs' alternative § 17510.6 charitable-solicitation theory and the Day 4 Birchall block: the master lease was held by **Musk Industries LLC** (`043026TT.txt:4078`) and Birchall, not OpenAI, decided which expenses got paid (DX 600: \"Elon holds 100 percent of the power and authority regarding the building\").\n- **Box upload:** 2026-05-02 09:41:20 PT — Saturday off-day batch.\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~445 KB, 2 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `1092.pdf`.\n\n## Transcribed text\n\n> **From:** Jared Birchall <jared@excession.com>\n> **To:** Chris Clark <chris@openai.com>\n> **Cc:** Shivon Zilis <shivon@neuralink.com>\n> **Subject:** Re: Quotes?\n> **Date:** 11/29/2017 5:52:55 PM\n>\n> Yes, as the profile of the companies grows and as E spends more time on site, we want to make sure we are erring on the side of caution.\n\n> **From:** Chris Clark <chris@openai.com>\n> **To:** Jared Birchall <jared@excession.com>\n> **Cc:** Shivon Zilis <shivon@neuralink.com>\n> **Subject:** Re: Quotes?\n> **Date:** Wed, Nov 29, 2017 at 1:53 AM\n>\n> Hi Jared,\n>\n> I don't think they included armed coverage but I will double check. Would you prefer armed coverage?\n>\n> Chris\n\n> **From:** Jared Birchall <jared@excession.com>\n> **To:** Chris Clark <chris@openai.com>\n> **Cc:** Shivon Zilis <shivon@neuralink.com>\n> **Subject:** Re: Quotes?\n> **Date:** Tue, Nov 28, 2017 at 3:43 PM\n>\n> Hey Chris,\n>\n> I hope you are well. Shivon shared the info below. Were any of these for armed coverage?\n>\n> Thanks,\n> Jared\n>\n> ---------- Forwarded message ----------\n\n> **From:** Chris Clark <chris@openai.com>\n> **To:** Shivon Zilis <shivon@openai.com>\n> **Subject:** Re: Quotes?\n> **Date:** Wed, Nov 22, 2017 at 3:38 PM\n>\n> Thanks! See below for the things we discussed:\n>\n> **Security Quotes:**\n>\n> *Bannerman*\n> - One of the most common providers for medium sized tech companies in the area\n> - 1 Professional Guard\n> - 7pm-8am M-F and 24/7 coverage on weekends (~113 hours a week)\n> - $3,250 per week\n> - Pros — Good reputation. Flexible scheduling. Solid tech features for relationship mgmt, scheduling, submitting inquiries, etc. Reasonably priced.\n> - Cons — While they try to assign the same guard over the long term, they sometimes rotate and have turnover.\n>\n> *Security Industry Specialists*\n> - Provided services to Stripe when they were here and were recommended by them\n> - 1 Guard, and 1 Site Supervisor assigned to the location.\n> - $7,250 a week for 24/7 coverage, or ~$5k a week for coverage similar to Bannerman's proposal\n> - Pros — Solid provider used by medium/large companies with good reviews and dedicated site supervisor\n> - Cons — Quite expensive. Typically only offer 24/7 services rather than after hours and weekend support.\n>\n> *Inter-State Security*\n> - Recommended by couple other small tech companies.\n> - Small scale provider\n> - $25 an hour / $3k per week\n> - Pros — Small company, easy to work with\n> - Cons — No tech or service mgmt features. Only real way to communicate/coordinate with their company is via phone.\n>\n> If we move forward I think my personal preference would be to give Bannerman a try, but I'm open to other suggestions.\n>\n> I think the easiest payment mechanism might be to just bump up the monthly building lease grant Jared already has in place by the estimated monthly security cost. That way you don't have to setup a new transfer or deal with reimbursement requests.\n>\n> **Building Access Controls**\n>\n> The other thing we really need to fix is the entry system. It's a hodgepodge of systems leftover from the Stripe days that are vulnerable to the most basic break in attempts. It also broke again today and the front door is stuck in an unlocked state until it's fixed later today. The proposed overhaul would include (1) new, more secure front door hardware and video doorbell/monitoring system a guard or receptionist can use to grant visitor access. It would also sync with the suite access systems for both NL and OpenAI suites. (2) We should also replace the parking gate control with a more secure access mechanism tied to everything else. Right now anyone with a basic remote and the right dip switch sequence can enter, and we have no way of de-credentialing fobs that are lost/stolen.\n>\n> The quote we have for the entry + parking gate upgrades is ~$25k, and the work would be done by MetroLock, which is the company that's been handling the building access systems since Stripe was here and seems to best understand who to unify everything.\n>\n> Let me know if I can answer any additional questions.\n>\n> Chris\n\n> **From:** Shivon Zilis <shivon@openai.com>\n> **To:** Chris Clark <chris@openai.com>\n> **Subject:** Quotes?\n> **Date:** Wed, Nov 22, 2017 at 10:58 AM\n>\n> For security? Can you send? Going to problem solve with Jared!\n>\n> Thank you thank you :)\n\n## Commentary\n\nDX 1092 is a granular operational document but it is dispositive on a small structural point that runs through plaintiffs' Pioneer Building theory: in November 2017, **Jared Birchall** — Musk's family-office CFO — was the financial decision-maker on **OpenAI/Neuralink shared-building security**, and the suggested payment mechanism was to \"**bump up the monthly building lease grant Jared already has in place**.\" That construct (Musk-side Excession/Musk Industries LLC making lease/security grants to OpenAI through Birchall) is the same architecture Birchall told Sam Teller in DX 600 (Jan 2017): \"**Elon holds 100 percent of the power and authority regarding the building**\" (`043026TT.txt:5829`). The exhibit also corroborates Clark's autonomy-deference (already shown in [[DX-668]]) — operationally OpenAI handled the vendor sourcing, but Birchall held veto/funding authority. Note Zilis's email signature on the Nov 22 message (`shivon@neuralink.com` on Birchall's reply, `shivon@openai.com` on the underlying thread) — a small contemporaneous fact about how blurred the Musk-orbit / OpenAI roles were in late 2017. Useful color for the [[Key Themes#The Pioneer Building|Pioneer Building]] thread but not a centerpiece exhibit; expect it to be cited only as one entry in the larger paper trail of Musk-side control of the building's operations.\n\n---\n*See also:* [[Day 4|Day 4 digest]] · [[Jared Birchall]] · [[Chris Clark]] · [[Shivon Zilis]] · [[DX-668]] · [[Pioneer Building]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-1145", "exhibit": "DX 1145", "party": "Defense", "type": "Tweet thread", "admitted_trial_day": "Day 8 (May 6, 2026)", "uploaded_box_pt": "2026-05-06T14:57:03", "uploader": "Morrison Foerster", "pages": 2, "size_bytes": 827673, "source_pdf": "DX-1145.pdf", "pdf_url": "https://media.mts-in.com/DX-1145.pdf", "body_markdown": "# DX-1145 — Feb 17, 2023 Musk tweet: \"closed source, maximum-profit company effectively controlled by Microsoft\"\n\n> Musk's February 17, 2023 quote-tweet declaring OpenAI had become \"a closed source, maximum-profit company effectively controlled by Microsoft\" — \"Not what I intended at all\" — defense's anchor for an SOL clock starting more than two years before the suit was filed.\n\n## Document type\n**Tweet thread** (two screenshots). Page 1: a Genevieve Roch-Decter (@GRDecter) tweet from 11:48 AM, Feb 15, 2023 (\"Elon Musk says that A.I. is 'one of the biggest risks' to civilization and needs to be regulated. He co-founded OpenAI\"). Page 2: Musk's @elonmusk reply at 4:36 AM, Feb 17, 2023 (14M views). Bates OPENAI_MUSK00039692–00039693.\n\n## Logistics\n- **Trial admission:** Day 8 (May 6, 2026) — Murati live; Zilis live; Toner video begins.\n- **Box upload:** 2026-05-06 14:57:03 PT — Day-8 batch.\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~828 KB, 2 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `1145.pdf`.\n\n## Transcribed text\n\n> **Genevieve Roch-Decter, CFA** (@GRDecter) [11:48 AM · Feb 15, 2023 · 11.6M Views]:\n>\n> Elon Musk says that A.I. is 'one of the biggest risks' to civilization and needs to be regulated\n>\n> He co-founded OpenAI\n>\n> ---\n>\n> **Elon Musk** (@elonmusk) [4:36 AM · Feb 17, 2023 · 14M Views]:\n>\n> OpenAI was created as an open source (which is why I named it \"Open\" AI), non-profit company to serve as a counterweight to Google, but now it has become **a closed source, maximum-profit company effectively controlled by Microsoft**.\n>\n> Not what I intended at all.\n\n## Commentary\n\nDX 1145 is the cleanest defense statute-of-limitations exhibit in this chunk. The February 17, 2023 tweet contains all four ingredients of plaintiffs' eventual breach theory — closed source, maximum profit, Microsoft control, \"not what I intended\" — articulated publicly to **14 million viewers** more than a year before the original Musk v. Altman complaint was filed. It pairs with the Sept. 24, 2020 \"OpenAI is essentially captured by Microsoft\" tweet ([[Key Themes]] SOL section, `042826TT.txt`/`042926TT.txt:2154`) and Musk's three-phases testimony — \"Phase three — they're looting the nonprofit. We're currently in phase three\" (`042926TT.txt:2133`). The defense uses it on [[Day 8|Day 8]] to box in plaintiffs' \"I concluded that the car had indeed been stolen … in 2023\" timing argument: if Musk publicly said it in February 2023, the clock started in February 2023. It also undercuts Musk's \"**The 'OPEN' in OpenAI represents open source**\" framing on direct (`042826TT.txt:5163`) by anchoring it to a specific public statement that triggers limitations.\n\n---\n\n*See also:* [[Day 8|Day 8 digest]] · [[Key Themes]] · [[Statute of Limitations]] · [[PX 251]]\n"}
{"exhibit_id": "DX-1156", "exhibit": "DX 1156", "party": "Defense", "type": "Text messages", "admitted_trial_day": "Day 4 (April 30, 2026)", "uploaded_box_pt": "2026-04-30T15:13:15", "uploader": "Morrison Foerster", "pages": 2, "size_bytes": 90555, "source_pdf": "DX-1156.pdf", "pdf_url": "https://media.mts-in.com/DX-1156.pdf", "body_markdown": "# DX 1156 — Feb 3, 2025 Musk → Zuckerberg \"bidding on the OpenAI IP\" texts\n\n> A two-page, 6-message text exchange on February 3, 2025 in which Elon Musk asks Mark Zuckerberg, \"**Are you open to the idea of bidding on the OpenAI IP with me and some others?**\" — and Zuckerberg replies \"**Want to discuss live?**\" before Musk says he will call in the morning.\n\n## Document type\n**Text messages, Cellebrite-style \"Short Message Report\" extraction.** Two pages: cover sheet (1 conversation, 6 messages, 3 participants — two Elon Musk handles plus Mark Zuckerberg, with handles fully redacted) and one page of message-by-message transcript with timestamps in GMT-08:00. Bates 2024MUSK-0006248–6249. Document ID DX-1156.\n\n## Logistics\n- **Trial admission:** Day 4 (April 30, 2026). Per [[Key Themes]]: \"DX 1156 — Feb 2025 Musk-Zuckerberg bidding-on-OpenAI-IP.\" Used by defense during Birchall direct/Musk cross to anchor the February 2025 Musk-led $97B bid for OpenAI's nonprofit assets.\n- **Box upload:** 2026-04-30 15:13:15 PT — Day 4 mid-afternoon batch (clustered with DX 539, DX 646, DX 1285).\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~88 KB, 2 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `1156 .pdf` (note trailing space in filename).\n\n## Transcribed text\n\n> **Conversation:** [chat — handle redacted]\n> **Date Range:** 2/3/2025\n> **Participants:** Elon Musk <[redacted]>, Elon Musk <[redacted]>, Mark Zuckerberg <[redacted]>\n>\n> **Messages in chronological order (GMT -08:00):**\n>\n> **EM** (Elon Musk) — 2/3/2025, 10:32 PM\n> Are you open to the idea of bidding on the OpenAI IP with me and some others?\n>\n> Receipts • Mark Zuckerberg [redacted] [R: 2/3/2025, 10:33 PM]\n>\n> **MZ** (Mark Zuckerberg) — 2/3/2025, 10:34 PM\n> Want to discuss live?\n>\n> Receipts • Elon Musk [redacted] [R: 2/3/2025, 10:59 PM]\n>\n> **EM** — 2/3/2025, 10:59 PM\n> Liked \"Want to discuss live?\"\n>\n> Receipts • Mark Zuckerberg [redacted] [D: 2/3/2025, 10:59 PM]\n>\n> **EM** — 2/3/2025, 10:59 PM\n> Will call in the morning\n>\n> Receipts • Mark Zuckerberg [redacted] [R: 2/4/2025, 3:09 PM]\n\n## Commentary\n\nDX 1156 is a tight, high-impact defense exhibit. The February 3, 2025 thread sits one week before Musk and a consortium of investors publicly announced an unsolicited **$97.4B bid** for the nonprofit's controlling assets in OpenAI Inc. — a bid the OpenAI board unanimously rejected. Defense uses this exhibit for two purposes: (i) to argue that Musk's stated charitable-trust theory is in tension with his attempt to **acquire** the nonprofit's assets in a competitive bid (i.e., if the assets belong \"to humanity,\" why is Musk trying to buy them with Zuckerberg?); and (ii) to undermine Musk's \"phase three\" disillusionment narrative by showing that the same person now suing for breach of the nonprofit mission was simultaneously trying to control the residual nonprofit IP. The Zuckerberg participation matters because it places a competing AI-company CEO on the bid syndicate — relevant to plaintiffs' competitive-harm theory (which defense argues cuts against Musk's mission framing). The \"**bidding on the OpenAI IP**\" phrasing — Musk's own words — is what the defense is likely to feature on cross. Cross-reference: [[PX 355]] (May 2023 \"de facto control of AGI\" texts) and [[PX 296]] (Oct 2022 \"bait and switch\") for the longer-arc Musk position.\n\n---\n*See also:* [[Day 4|Day 4 digest]] · [[Sam Altman]] · [[PX 355]] · [[PX 296]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-1157", "exhibit": "DX 1157", "party": "Defense", "type": "Letter (letter of intent)", "admitted_trial_day": "Day 4 (April 30, 2026)", "uploaded_box_pt": "2026-04-30T15:13:20", "uploader": "Morrison Foerster", "pages": 4, "size_bytes": 1652920, "source_pdf": "DX-1157.pdf", "pdf_url": "https://media.mts-in.com/DX-1157.pdf", "body_markdown": "# DX 1157 — Feb 10, 2025 Toberoff $97.375B Letter of Intent to acquire all assets of OpenAI, Inc.\n\n> The signed February 10, 2025 Letter of Intent on Toberoff & Associates letterhead, addressed to OpenAI's eight directors, in which a buyer consortium led by **X.AI Corp.** (Elon Musk, CEO) — and including Baron Capital, Valor Management, Atreides Management, Vy Fund III, Emanuel Capital Management, and Eight Partners VC — proposed to acquire **all assets** of OpenAI, Inc. (including the \"OpenAI For-Profit Entities\") for **$97,375,000,000 in cash**.\n\n## Document type\n**Court filing / Letter (Letter of Intent).** Four pages: page 1 cover (Toberoff & Associates letterhead, recipient list — Bret Taylor, Sam Altman, Adam D'Angelo, Sue Desmond-Hellmann, Zico Kolter, Adebayo Ogunlesi, Paul Nakasone, Nicole Seligman, Fidji Simo, Larry Summers); page 2 standard LOI provisions (definitive agreement, due diligence, expenses, term, governing law — **Texas / W.D. Tex. Austin**); page 3 signature block (Marc Toberoff plus signed counterparts from Baron Capital (Ron Baron), X.AI Corp. (**Elon Musk, CEO**), Valor Management (Antonio Gracias), Atreides Management (Gavin Baker), Vy Fund III (George Bashforth), Emanuel Capital Management (Ariel Z. Emanuel), Eight Partners VC (Ian M. Shannon)); page 4 Exhibit A — purchase price and terms. Bates OPENAI_MUSK00038583–38586. Document ID DX-1157.\n\n## Logistics\n- **Trial admission:** Day 4 (April 30, 2026). Pre-trial wiki reference: **\"DX 1157 — Feb 10 2025 Toberoff $97B bid letter.\"** Used during defense Birchall direct / Musk cross — the formal counterpart to [[DX 1156]] (the same week's Musk-Zuckerberg texts).\n- **Box upload:** 2026-04-30 15:13:20 PT — Day 4 mid-afternoon defense batch (clustered with [[DX 502]], [[DX 545]], [[DX 1156]]).\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~1.58 MB, 4 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `1157 .pdf` (note trailing space).\n\n## Transcribed text\n\n> **TOBEROFF & Associates, P.C.**\n> 23823 Malibu Road, Suite 50-363, Malibu, CA 90265\n> Telephone: +1 310 246 3333\n> *Marc Toberoff* — Direct: +1 310 736 5666 — mtoberoff@toberoffandassociates.com\n>\n> February 10, 2025\n>\n> *Via Email: WDSavitt@wlrk.com; SKEddy@wlrk.com*\n>\n> Bret Taylor, Chairperson · Samuel Altman, CEO · Adam D'Angelo · Dr. Sue Desmond-Hellmann · Zico Kolter\n> Adebayo Ogunlesi · Gen. Paul Nakasone · Nicole Seligman · Fidji Simo · Larry Summers\n>\n> OpenAI, Inc.\n> 550 Terry A Francois Blvd.\n> San Francisco, CA 94158\n>\n> Dear Members of the Board of OpenAI, Inc.:\n>\n> On behalf of a consortium of investors (the \"Buyer\"), including **X.AI Corp.**, Baron Capital Group, Inc., Valor Management LLC, Atreides Management, LP, Vy Fund III, L.P., Emanuel Capital Management, LLC, Eight Partners VC, LLC, and others, I am pleased to submit this letter of intent (\"Letter of Intent\") outlining their proposal to acquire **all assets** (the \"Business\") of OpenAI, Inc. (the \"Company\"), on substantially the proposed terms and conditions set forth in Exhibit A hereto (the \"Transaction\"). This proposal is subject in all respects to the terms and conditions of this Letter of Intent.\n>\n> 1. **Definitive Agreement.** The obligations of Buyer, the Company, and its Members to consummate the Transaction and related transactions are subject to and conditioned upon (a) the negotiation and execution by Company and Buyer of a mutually acceptable definitive agreement…\n>\n> 2. **Due Diligence.** During the Term of this Letter of Intent, the Company will, upon reasonable advance notice and during normal business hours, (a) afford Buyer and its representatives customary due diligence investigation of the Company and its Business including access to any assets, facilities, equipment, books, and records of the Company in connection with the Business, and (b) upon reasonable request from Buyer with respect to particular personnel involved in the Business, grant Buyer and its representatives access to such personnel.\n>\n> 3. **Operation of the Business.** … Company shall use reasonable efforts to continue to operate the Business in the ordinary course consistent with past practices.\n>\n> 4. **Expenses.** … each of the parties hereto will pay its own costs and expenses…\n>\n> 5. **Counterparts.**\n>\n> 6. **Integration; Modifications.**\n>\n> 7. **Term.** … will terminate upon the earliest to occur of … (b) **May 10, 2025**; (c) the date that the parties mutually agree in writing to terminate this Letter of Intent; or (d) upon the Buyer's receipt of the Company's written declination to negotiate and/or enter into a Definitive Agreement.\n>\n> 8. **Governing Law; Consent to Jurisdiction.** This Letter of Intent … will be governed by and construed under and in accordance with the laws of the **State of Texas** … exclusive jurisdiction of the United States District Court for the **Western District of Texas**, or … any Texas State court, sitting in **Austin, Texas** …\n>\n> Very truly yours,\n> /s/ Marc Toberoff\n>\n> [Counterpart signatures, page 3:]\n> - **Baron Capital Group, Inc.** — Ron Baron, CEO\n> - **X.AI, Corp.** — **Elon Musk, CEO**\n> - **Valor Management LLC** — Antonio Gracias, CEO and CIO\n> - **Atreides Management, LP** — Gavin Baker, Managing Partner & CIO\n> - **Vy Fund III, L.P.** — George Bashforth, Director\n> - **Emanuel Capital Management, LLC** — Ariel Z. Emanuel, CEO\n> - **Eight Partners VC, LLC** — Ian M. Shannon, General Counsel\n>\n> ACCEPTED AND AGREED:\n> **OpenAI, Inc.** — By: [blank] — Name: [blank] — Title: [blank] *(unsigned by OpenAI)*\n>\n> ---\n>\n> **Exhibit A — Principal Terms and Conditions**\n>\n> 1. **Purchase Price and Terms.** \"Based upon Buyer's limited review of selected financial and other information available to Buyer to date, Buyer would expect to acquire **all assets of the Company for approximately Ninety-Seven Billion Three Hundred Seventy-Five Million U.S. Dollars ($97,375,000,000)**. Buyer's valuation is based on the Company's historical financial results as well as projections for the Company available to Buyer to date, and it assumes the Company's public representations as to its ownership and complete control of its various subsidiary or affiliated for-profit entities (the '**OpenAI For-Profit Entities**') are accurate. Buyer anticipates that **one hundred percent (100%) of the total purchase price would be paid in cash**. The structure of the Transaction will be determined after taking into further account tax and other legal considerations.\"\n>\n> 2. **Source of Financing.** \"Buyer will not require third-party debt financing to be a contingency to closing the Transaction.\"\n>\n> 3. **Due Diligence.** \"Buyer's proposal is subject to customary due diligence investigation of the Company, including of the OpenAI For-Profit Entities, satisfactory to Buyer in its sole discretion. Key due diligence activities would include, among other things, (i) a complete legal, financial (including an audit), and tax review of the Company and its OpenAI For-Profit Entities, (ii) a detailed review of the Company's financial projections, including as to the OpenAI For-Profit Entities, with a focus on the key drivers of revenue growth and EBITDA, and (iii) detailed discussions with management.\"\n>\n> 4. **Definitive Documentation.** Customary representations, warranties, conditions, covenants and indemnification.\n>\n> 5. **Time and Process.** \"Buyer's expectation would be to close within thirty (30) days of the completion of its due diligence.\"\n\n## Commentary\n\nDX 1157 is the documentary cornerstone of the defense's [[Key Themes#\"Pageant of hypocrisy\" / Tale of two Elons|\"pageant of hypocrisy\"]] theme. Plaintiffs' theory is that OpenAI's assets belong to a charitable trust whose mission was **for humanity** — and Musk's standing depends on framing the for-profit conversion as looting. **Three weeks before this trial began**, Musk himself signed (as **CEO of X.AI Corp.**) a $97.375 billion all-cash offer to **acquire 100% of those same assets**, alongside Mark Zuckerberg-aligned (per [[DX 1156]]) and his own xAI consortium. Defense uses the LOI to make four points: (i) Musk's competing AI company (xAI / X.AI Corp.) — not the Musk Foundation, not a charitable vehicle — is the named buyer, contradicting the \"non-profit charitable interest\" framing; (ii) the **purchase price formula assumes \"the Company's public representations as to its ownership and complete control of its various subsidiary or affiliated for-profit entities are accurate\"** — i.e., Musk *valued the OpenAI assets on the basis of the very for-profit structure he is suing over*; (iii) the choice of **Texas jurisdiction** (Austin, where xAI is headquartered) is read as Musk seeking a friendly forum; and (iv) the OpenAI board declined the bid (the \"ACCEPTED AND AGREED\" line is unsigned), and the bid expired by its own terms on **May 10, 2025** — making the lawsuit the alternative path to the same prize. The buyer-consortium signatures are themselves notable: Antonio Gracias, Gavin Baker, Ron Baron, and Ari Emanuel are all longtime Musk associates. Cross-reference: [[DX 1156]] (Feb. 3, 2025 Musk-Zuckerberg \"bidding on the OpenAI IP\" texts) — the same week's private architecture for the same bid.\n\n---\n*See also:* [[Day 4|Day 4 digest]] · [[Sam Altman]] · [[Jared Birchall]] · [[DX 1156]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-1201", "exhibit": "DX 1201", "party": "Defendants", "type": "Demonstrative", "admitted_trial_day": "Day 6 (May 4, 2026)", "uploaded_box_pt": "2026-05-04T14:32:18", "uploader": "Morrison Foerster", "pages": 1, "size_bytes": 844288, "source_pdf": "DX-1201.pdf", "pdf_url": "https://media.mts-in.com/DX-1201.pdf", "body_markdown": "# DX 1201 — Photograph: Early OpenAI working session in a residential setting (likely Brockman's apartment, 2015–2016)\n\n> A single photograph admitted as a Day 6 demonstrative during Russell direct / Brockman cross. The image shows roughly six people at a long table in what appears to be a residential loft / apartment — exposed wood-beam ceiling, hardwood floors, a TV with attached speakers, framed art, a houseplant — with laptops, notebooks, coffee cups, a backpack on the table, and a large flip-pad / poster of handwritten diagrams in the center. Used by defense as a piece of foundational color about the **scrappy, household-scale origins** of OpenAI in 2015–2016, before any of the for-profit conversion plans were on the table.\n\n## Document type\n**Demonstrative — photograph.** Color photo, single page, no caption or annotations on the produced image. Bates OPENAI_MUSK00039802.\n\n## Logistics\n- **Trial admission:** Day 6 (May 4, 2026) — Russell direct + Brockman cross. Used as a defense demonstrative during Brockman cross to establish the founding-era working environment — six people, a notepad, a residential setting — against which plaintiffs' \"premeditated billion-dollar takeover\" theory has to be measured.\n- **Box upload:** 2026-05-04 14:32:18 PT — Day 6 mid-afternoon batch.\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~824 KB, 1 page.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `1201.pdf`.\n\n## Transcribed text\n\n> *[No transcribable text. The exhibit consists of a single photograph showing approximately six people at a long dark table in a residential loft-style room. Visible details include exposed dark-stained wooden ceiling beams against white plaster, three pendant lights, hardwood flooring, a flat-screen TV with floor-standing speakers along the back wall, two framed pieces of artwork on the side walls, a tall houseplant in the back-right corner, an upholstered grey sofa with a person seated on it watching the others, and the working table itself bearing several open laptops, a coffee cup, a glass of water, a blue backpack, notebooks, and a large white poster / sketchpad displaying handwritten diagrams positioned at the table's center. Two participants in the foreground are visible in profile; one wears a dark \"Toronto\" sweatshirt; another at far right wears a red shirt and appears to be drinking from a mug. Lighting is low and warm, suggesting an evening session.]*\n\n## Commentary\n\nDX 1201 is, in form, a single photograph — but the defense uses it for narrative work that connects to the case's most contested factual question: **what did the founders actually intend OpenAI to be in 2015–2016?** **Defense's read:** the room is a residential loft, the table is a kitchen-scale working surface, and the team is six people — the visual antithesis of a premeditated $300B+ corporate takeover. The \"Toronto\" sweatshirt is consistent with [[Ilya Sutskever]] (University of Toronto Ph.D., 2013) or one of the other early Toronto-pipeline researchers ([[Wojciech Zaremba]], [[Andrej Karpathy]], etc.), and the residential setting is consistent with the documented early use of [[Greg Brockman]]'s San Francisco apartment as an OpenAI working space prior to the lease of the [[Pioneer Building]] (see [[Key Themes]] § \"The Pioneer Building\"). The image — admitted as a demonstrative — is the kind of evidence that pairs naturally with [[Brockman Journal]] entries about the early team and with the **\"AI research non-profit (through end of 2017)\"** framing in [[DX 631]]. **Plaintiffs' read:** the photograph proves nothing about intent at the time of incorporation — visual modesty in 2015–2016 is fully compatible with plaintiffs' theory that the for-profit pivot was *contemplated from inception* but operationally deferred until the team's research had produced a marketable AGI-trajectory asset (see [[PX 156]] / [[PX 157]] and [[Brockman Journal]] for the August–September 2017 \"this is the only chance we have to get out from under Elon\" sequence). Demonstratives like this one risk being treated by juries as proof of \"humble beginnings = no bad intent,\" which is a fallacy plaintiffs argued against in the Day 6 sidebar before admission. The exhibit was admitted with the caveat that the photograph is offered solely as illustrative of the early working environment and not as substantive evidence of any factual claim about corporate structure or charitable intent.\n\n---\n*See also:* [[Day 6|Day 6 digest]] · [[Greg Brockman]] · [[Ilya Sutskever]] · [[Pioneer Building]] · [[Brockman Journal]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-1203", "exhibit": "DX 1203", "party": "Defense (OpenAI/MS)", "type": "Demonstrative", "admitted_trial_day": "Day 7 (May 5, 2026)", "uploaded_box_pt": "2026-05-05T14:27:02", "uploader": "OpenAI startup-fund defense firm", "pages": 1, "size_bytes": 1166689, "source_pdf": "DX-1203.pdf", "pdf_url": "https://media.mts-in.com/DX-1203.pdf", "body_markdown": "# DX 1203 — Brockman group selfie with OpenAI employees\n\n> A single full-page color photograph: Greg Brockman in the foreground taking a selfie with a large, smiling group of OpenAI employees in an office common area — the defense's humanizing demonstrative used on Brockman's Day 7 direct.\n\n## Document type\n**Demonstrative (photograph).** One-page color image, no metadata captions, no Bates beyond production stamp OPENAI_MUSK00039804.\n\n## Logistics\n- **Trial admission:** Day 7 (May 5, 2026) — Brockman defense direct + Mr. Molo recross + DX 1252; Wu deposition.\n- **Box upload:** 2026-05-05 14:27:02 PT — same-day upload during the Brockman direct examination.\n- **Uploader:** OpenAI startup-fund defense firm.\n- **File size:** ~1.1 MB, 1 page.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `1203.pdf`.\n\n## Transcribed text\n\n> [Photograph only. No transcribable text. Caption from court stamp: \"United States District Court, Northern District of California — Case No. 4:24-CV-04722-YGR — Musk et al. v. Altman et al. — Exhibit No. DX-1203 — Mark Busby, Clerk.\" Image content: Greg Brockman, in a black leather jacket, taking a forward-facing selfie at an OpenAI office; ~50–60 employees fill the frame behind him, many waving and smiling, some flashing peace signs. Indoor office setting with paper-globe pendant lights and a wall clock visible. CONFIDENTIAL stamp absent.]\n\n## Commentary\n\nDX 1203 is purely demonstrative — defense counsel used it on Brockman's direct to cast him as the embedded operational leader of a team that loved him, reinforcing the \"**fire Elon**\" frame defense had built through DX 1252 the same day (see [[Key Themes]] \"**Real decision is fire Elon**\"). It is the visual counterpoint to Mr. Molo's recross frame — \"**It's just an after-the-fact manufactured excuse for the very honest emotion that you expressed in the journal at the time**\" — and to plaintiffs' \"**$30 billion**\" set piece on Day 6 ([[Brockman Journal]]). The image carries no probative weight on the charitable-trust elements, but it is the kind of human-warmth artifact that defense placed on the screen during Brockman's direct to soften the journal-reading sequence the jury had just absorbed.\n\n---\n*See also:* [[Day 7|Day 7 digest]] · [[Greg Brockman]] · [[Brockman Journal]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-1204", "exhibit": "DX 1204", "party": "Defense", "type": "Demonstrative", "admitted_trial_day": "Day 7 (May 5, 2026)", "uploaded_box_pt": "2026-05-05T14:27:02", "uploader": "Morrison Foerster", "pages": 1, "size_bytes": 1011022, "source_pdf": "DX-1204.pdf", "pdf_url": "https://media.mts-in.com/DX-1204.pdf", "body_markdown": "# DX 1204 — Group selfie photograph (OpenAI conference room) — Brockman defense direct demonstrative\n\n> A single high-resolution **group selfie photograph** taken in an OpenAI conference room — used as a defense demonstrative on Day 7 during Brockman's defense direct, almost certainly to humanize Brockman (the photographer/foreground subject) and to show the live OpenAI team in their workspace.\n\n## Document type\n**Demonstrative** — single-page color photograph taken in a glass-walled OpenAI conference room. The photo shows ~20 people seated around or standing behind a long wooden conference table, a projector beam visible mid-image, laptops open on the table. The photographer (foreground left) is the most prominent figure; identification of individual people in the photograph would require pairing with the Day 7 transcript at the moment the exhibit was published. Bates OPENAI_MUSK00039805. The PDF carries a DX-1204 Defendant exhibit stamp.\n\n## Logistics\n- **Trial admission:** Day 7 (May 5, 2026) — Brockman defense direct + Mr. Molo recross + DX 1252; Wu deposition. The single-page photograph is consistent with a \"this is the team\" / \"this is the workspace\" demonstrative used during a witness's biographical or scene-setting direct.\n- **Box upload:** 2026-05-05 14:27:02 PT — Day 7 mid-afternoon Defense batch (uploaded same time as [[DX 899]]).\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~988 KB, 1 page.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `1204.pdf`.\n\n## Transcribed text\n\n> [No textual content — this exhibit is a single full-page color photograph. Verbal description: a glass-walled OpenAI conference room; long wooden conference table extending into the background; ~20 people in the frame, several waving or smiling at the camera; a ceiling-mounted projector visible mid-image; multiple open laptops on the table; large pendant lamps along the back wall; the photographer (foreground left) appears wearing a black leather jacket and is the most prominent subject. The photo is consistent with a posed group selfie taken during or just after a meeting.]\n\n## Commentary\n\nDX 1204 is a **defense demonstrative**, almost certainly used during Brockman's Day 7 defense direct to humanize the witness and the OpenAI team — the standard rhetorical move to put faces on the institution that plaintiffs have spent six trial days describing as a charity-stealing operation (see [[Key Themes]] §\"Pageant of hypocrisy / Tale of two Elons\" for the defense's framing). Without the Day 7 transcript reference at the publication moment, individual identification would be speculation. The photograph's role in the trial record is best understood as scene-setting around Brockman's direct — pair with the substantive Day 7 documents [[DX 899]] (the June 2020 OpenAI API blog) and DX 1252 (the November 5, 2017 Brockman journal entry introduced over Mr. Molo's hearsay objection as a prior consistent statement, see [[Brockman Journal]] and [[Key Themes]] §\"DX 1252\").\n\n---\n*See also:* [[Day 7|Day 7 digest]] · [[DX 899]] · [[Greg Brockman]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-1205", "exhibit": "DX 1205", "party": "Defense", "type": "Demonstrative", "admitted_trial_day": "Day 7 (May 5, 2026)", "uploaded_box_pt": "2026-05-05T14:27:02", "uploader": "Morrison Foerster", "pages": 1, "size_bytes": 858125, "source_pdf": "DX-1205.pdf", "pdf_url": "https://media.mts-in.com/DX-1205.pdf", "body_markdown": "# DX 1205 — Photograph of Greg Brockman with Ilya Sutskever (Sutskever's animal-painting wall)\n\n> A single full-page color photograph, used as a defense direct demonstrative on Day 7 — two men standing arm-around-shoulders, smiling, in front of a wall of Ilya Sutskever's hand-painted animal canvases (tiger, hippo, sloth, hippo-on-yellow, cat-on-yellow, elephants, lion, etc.).\n\n## Document type\n**Demonstrative — single photograph.** One color image filling the page, with the standard DX exhibit-stamp box in the bottom-left corner (\"Δ DEFENDANT Δ … DX-1205\"). Bates OPENAI_MUSK00039806. By the visual context — Sutskever's well-documented animal paintings — the second figure is almost certainly Ilya Sutskever; the leather-jacketed figure is consistent with Greg Brockman's appearance on the stand.\n\n## Logistics\n- **Trial admission:** Day 7 (May 5, 2026 — Brockman defense direct + Mr. Molo recross + DX 1252; Wu deposition). Used during the Brockman defense direct as a humanizing personal-relationship demonstrative — i.e., the friendship-and-shared-mission framing that supports the [[Brockman Journal|\"morally bankrupt\"]] passage being read as resistance to *removing-Musk* rather than nonprofit-conversion intent.\n- **Box upload:** 2026-05-05 14:27:02 PT — Day 7 early-afternoon defense batch (uploaded two seconds after [[DX-704]]).\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~838 KB, 1 page.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `1205.pdf`.\n\n## Transcribed text\n\n[The exhibit is a single photograph — no text content.]\n\n> **Image:** Two men, arm-around-shoulders, smiling at the camera, standing in front of a white wall hung with ~10 hand-painted animal canvases (a tiger, hippo, sloth, hippopotamus on yellow, cat on yellow, elephants, lion, owl-or-similar, fish, and a rose). Floor and a brown leather couch are visible at the bottom. The composition is consistent with a photograph taken in someone's home or office.\n\n## Commentary\n\nDX 1205 is a defense demonstrative whose function is **emotional-register correction** rather than factual proof. After two days of Brockman cross built around the [[Brockman Journal|\"only chance to get out from Elon\" / \"morally bankrupt\" / \"lie\"]] passages, the defense's direct of Brockman on Day 7 needed to put back into the room the human relationship between Brockman and Sutskever — the same Sutskever whose [[PX 98|January 1, 2018 \"most overwhelmingly competent person in the world\" note]] had been on the screen the week before, and the same Sutskever who would, in November 2023, lead the board vote to fire Sam Altman (see [[PX 309]]). The animal-painting wall is the visual texture that makes Sutskever a person rather than a transcript citation. This is a small exhibit. It is also the kind of exhibit that, in summation, becomes the image jurors associate with the founders' \"**we wanted to keep working together**\" framing — defense's answer to the [[Key Themes|$30 billion / paid-zero]] cross-examination set piece.\n\n---\n*See also:* [[Day 7|Day 7 digest]] · [[Greg Brockman]] · [[Ilya Sutskever]] · [[Brockman Journal]] · [[PX 98]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-1256", "exhibit": "DX 1256", "party": "Defense", "type": "Compendium / multi-doc", "admitted_trial_day": "Day 4 (April 30, 2026)", "uploaded_box_pt": "2026-04-30T15:13:31", "uploader": "Morrison Foerster", "pages": 282, "size_bytes": 17753947, "source_pdf": "DX-1256.pdf", "pdf_url": "https://media.mts-in.com/DX-1256.pdf", "body_markdown": "# DX 1256 — Fidelity Charitable Policy Guidelines compendium (2017–2020)\n\n> A 282-page compendium of seven successive editions of the **Fidelity Charitable Program Circular / Policy Guidelines** (last-updated July 2017, March 2018, June 2018, July 2018, October 2018, August 2019, March 2020, August 2020) — the governing document of the donor-advised fund used by Musk's Fidelity Charitable Giving Account, the program disclosures the defense relies on for the [[Key Themes#The \"donor-advised funds\" / tax-deduction wrinkle|DAF / tax-deduction wrinkle]].\n\n## Document type\n**Compendium of public disclosure documents (program circulars), 282 pages.** Eight chronologically-stacked editions of Fidelity Charitable's Policy Guidelines / Program Circular, each covering: Public Charity / DAF program overview, Establishing a Giving Account, Contributions, Tax Considerations, Investment Options, Pool Allocation Process, Grantmaking, Recordkeeping & Reporting, Fees and Expenses, Successor Options (Individual / Charitable Organization / Endowed Giving Program), Trustees' Initiative, Service Providers, Glossary. Bates FIDCHAR-OPENAI-000128 through FIDCHAR-OPENAI-000401 (filed as Dkt. 328-27 through 328-34). Document ID DX-1256. Cover page identifies it simply as **\"Fidelity Charitable Policy Guidelines (2017–2022).\"**\n\n## Logistics\n- **Trial admission:** Day 4 (April 30, 2026) — Musk cross + redirect; Birchall direct + cross. Companion exhibit to [[DX-621]] (Musk's June 21, 2017 Fidelity Charitable application establishing **The Musk Foundation Charitable Fund**), and to plaintiffs' / defense's exchange around the [[Key Themes#The \"donor-advised funds\" / tax-deduction wrinkle|donor-advised-fund / tax-deduction wrinkle]] line of cross (Wilson to Birchall, `043026TT.txt:5413`).\n- **Box upload:** 2026-04-30 15:13:31 PT — Day 4 mid-afternoon batch (clustered with DX 621 and DX 668).\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~16.9 MB, 282 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `1256.pdf`. Underlying source is Fidelity Charitable's own publicly distributed Program Circular, produced in discovery from Fidelity Charitable's files.\n\n## Transcribed text\n\nThe document is a compendium of eight successive editions of Fidelity Charitable's standard public **Program Circular / Policy Guidelines**. Rather than reproduce every word of all eight versions (each ~30–35 pages of substantively similar boilerplate), key passages substantively identical across all editions are excerpted below, with material deltas noted.\n\n> **Title page:** \"Fidelity Charitable Policy Guidelines (2017–2022)\"\n>\n> ---\n>\n> **Public-charity status (all editions):**\n>\n> \"Fidelity Charitable, an independent, section 501(c)(3) public charity that administers donor-advised funds, was organized, and operates exclusively, for charitable purposes. Fidelity Charitable provides donors the programs, tools, resources, and support to make charitable giving accessible, simple, and effective.\"\n>\n> ---\n>\n> **Contributions are irrevocable (all editions, key text):**\n>\n> \"All contributions are subject to review and approval by the Trustees. The Trustees reserve the right, in their sole discretion and for any reason, not to accept any contribution. … Once Fidelity Charitable accepts a contribution, **it is irrevocable and is owned and controlled by the Trustees.** The Trustees have **exclusive legal control** over all contributed assets. **Contributions to Fidelity Charitable are not refundable.**\"\n>\n> ---\n>\n> **Tax treatment (all editions):**\n>\n> \"Donors are generally eligible to take an itemized deduction on the date the charitable contribution to Fidelity Charitable is made. … Deductions for contributions of long-term capital gain property (such as appreciated securities held for more than one year) may be taken **up to 30% of AGI**. Deductions for all other contributions (including contributions of short-term capital gain property and cash equivalents) may be taken **up to 50% of AGI**\" (raised to **60%** in 2018 and later editions).\n>\n> \"Any income that accrues to a Giving Account that is related to a contribution is income of Fidelity Charitable, **not of the Account Holder's taxable estate.** Therefore, the Account Holder is neither subject to tax on that income nor eligible to take further charitable contribution deductions with respect to that income.\"\n>\n> \"When Fidelity Charitable disburses grants to charities based on recommendations, **Fidelity Charitable is granting its own assets.** Accordingly, Account Holders who make grant recommendations are not eligible for additional charitable deductions for these grants.\"\n>\n> ---\n>\n> **Grant guidelines (substantively unchanged across editions):**\n>\n> \"Fidelity Charitable approves only those grants that are used **exclusively in furtherance of charitable purposes.** … Fidelity Charitable reserves the right … to decline to make a recommended grant to a charitable organization, including, without limitation: (i) where the grant will confer a more than incidental benefit on an Account Holder…; (ii) where the grant will be used for lobbying, for political contributions, or to support political campaign activities; (iii) where the grant will be used for improper purposes; (iv) where the Account Holder and related persons control the organization; (v) where Fidelity Charitable provides a substantial portion of the organization's public support; … (vi) [2020 edition adds] where the organization's governing board is composed entirely of related parties, or consists of fewer than three individuals.\"\n>\n> **More than incidental benefit (\"enforceable pledges\"):** \"Grants cannot be made that provide '**more than incidental benefits**' to a donor, other Account Holders, their families, or other third parties.\" Examples listed: tickets to charitable events, raffle tickets, membership benefits with goods/services, athletic-fund seating benefits, **financial obligations such as enforceable pledges**, school tuition, scholarships in which the donor has discretion. Lobbying / political-campaign grants are categorically prohibited.\n>\n> ---\n>\n> **Fees (consistent across editions, with minor adjustments):**\n>\n> Annual administrative fee: **0.60%** of Giving Account assets or $100 ($500 for corporate Giving Accounts), whichever is greater, charged to the investment pool on a daily basis. Tiered reductions for accounts above $500K (down to 15 bps for $2.5M-5M) and a flat schedule for $5M+ accounts (19 bps at $5M down to ~11.5 bps at $75M+). Brokerage commission schedule: 1.2¢/share standard, 1.7¢/share for large block or thinly traded.\n>\n> ---\n>\n> **Successor Options (all editions):** Three options — (1) Individual Successor(s); (2) Charitable Organization Successor(s); (3) Endowed Giving Program (recurring grants to up to six IRS-qualified public charities, minimum $100K balance at activation, 5%/year minimum distribution, five-year minimum duration).\n>\n> ---\n>\n> *[Remainder consists of substantially identical operational disclosures across the eight editions: contribution-processing timelines, EFT mechanics, securities-contribution mechanics, investment-pool descriptions (Aggressive Growth / Growth / Moderate Growth / Balanced / Moderate Income / Income / Conservative Income / various Single-Asset-Class, Sustainable & Impact, Charitable Legacy, Charitable Investment Advisor Program, Charitable DonorFlex Program), pool-allocation mechanics, grant-processing mechanics including the Gift4Giving® eGift program, Trustees' Initiative direct-grantmaking program (>$25M granted as of June 2019), service-provider disclosures (FMR LLC, Strategic Advisers LLC, National Charitable Services Corp., Fidelity Brokerage Services LLC), conflict-of-terms / limitation-of-liability boilerplate, and a glossary. The eight editions track minor program evolutions: addition of the Sustainable and Impact Investing Pools (2018), AGI-percentage update from 50% to 60% (2018 TCJA conformance), tightening of the inactivity policy from seven years to two years (2018→2020), and addition of governing-board-composition restrictions on grant recipients (2020).]*\n\n## Commentary\n\nDX 1256 is the **legal infrastructure document** for the [[Key Themes#The \"donor-advised funds\" / tax-deduction wrinkle|defense's donor-advised-fund / tax-deduction line]] — the eight successive editions of the Fidelity Charitable Program Circular under which Musk made a substantial portion of the donations plaintiffs claim were \"stolen\" by OpenAI. The four key facts the exhibit establishes for defense are: **(i)** every contribution Musk made to a Fidelity Charitable DAF was, by the document's own terms, \"**irrevocable and ... owned and controlled by the Trustees**\" with \"**exclusive legal control**\" over the assets — supporting the judge's Day 5 observation that restitution under the DAF-routed donations would have to \"**[go] back to the plaintiff**\" (`050126TT.txt:1446`) only with difficulty; **(ii)** Musk took an immediate itemized tax deduction at the moment of contribution to Fidelity Charitable, not at the moment of grant to OpenAI — which is the predicate for Wilson's \"for every single one of those donations, it's somewhere in the chain, Mr. Musk was entitled to take a tax deduction\" cross of Birchall (`043026TT.txt:5413`); **(iii)** at the grant-recommendation stage, \"**Fidelity Charitable is granting its own assets**\" — a structural fact plaintiffs must explain around; and **(iv)** the donor-advised-fund structure prohibits grants conferring \"**more than incidental benefit**\" on the donor, prohibits enforceable-pledge satisfaction, and reserves the right to decline grants where the donor and related persons control the recipient organization — all of which speak to whether Musk could plausibly have understood OpenAI as something he \"controlled\" or whose assets he was \"donating with strings attached.\" Birchall's concession on cross — \"**I'm not familiar with all of the legalities of donor advised funds**\" (`043026TT.txt:5007`) — is the wedge plaintiffs use to suggest the Musk side did not actually internalize this structure when making contributions. The 17.7 MB / 282-page size of the exhibit reflects defense counsel's apparent decision to file *all* relevant editions covering the period of Musk's giving rather than rely on a single version, foreclosing any argument that a particular term was added later.\n\n---\n*See also:* [[Day 4|Day 4 digest]] · [[DX-621]] · [[Jared Birchall]] · [[Musk Foundation]] · [[Donor-Advised Funds]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-1284", "exhibit": "DX 1284", "party": "Defense", "type": "Contract", "admitted_trial_day": "Day 4 (April 30, 2026)", "uploaded_box_pt": "2026-04-30T15:13:20", "uploader": "Morrison Foerster", "pages": 8, "size_bytes": 938972, "source_pdf": "DX-1284.pdf", "pdf_url": "https://media.mts-in.com/DX-1284.pdf", "body_markdown": "# DX-1284 — July 2014 Vanguard Charitable Gift Agreement establishing \"The Musk Charitable Fund\" (DAF)\n\n> The Vanguard Charitable Endowment Program \"Gift agreement for individuals or trusts\" creating Musk's donor-advised fund — the vehicle through which the bulk of Musk's OpenAI contributions were channeled, signed by Musk on 7-09-2014 and faxed 7-10-2014.\n\n## Document type\n**Contract** — Vanguard Charitable's standard 8-page Gift Agreement form for individuals or trusts. Establishes \"The Musk Charitable Fund\" as the donor-advised account name; donor is the **Elon Musk Revocable Trust** (trust agreement dated 07-22-2003); Owner/Trustee A is Elon R. Musk (2200 Geng Road, Suite 100, Palo Alto, CA 94303); Account Advisor B is Ronald F. Gong, Financial Advisor (myCFO); Successor-Advisor A is Antonio Gracias. Signed by Musk on 07-09-2014; Account Advisor signature dated 07-08-2014; faxed 07-10-2014, 8:18 from 888-426-3273. Bates VC000411–VC000418.\n\n## Logistics\n- **Trial admission:** Day 4 (April 30, 2026) — Musk cross + redirect; Birchall direct + cross.\n- **Box upload:** 2026-04-30 15:13:20 PT — Day-4 mid-afternoon batch (clustered with DX 827, DX 857).\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~939 KB, 8 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `1284.pdf`.\n\n## Transcribed text\n\n> **VANGUARD CHARITABLE — Gift Agreement for individuals or trusts**\n>\n> \"Establish a philanthropic account with your irrevocable gift of at least $25,000 in value.\"\n>\n> **1. Account Name:** The **Musk Charitable Fund**\n>\n> **2. Donor Information** — *Trust:*\n> - Name of Trust: **Elon Musk Revocable Trust**\n> - Trust Agreement Date: **07-22-2003**\n> - Owner/Trustee A: **Mr. Elon R. Musk**, 2200 Geng Road, Suite 100, Palo Alto, CA 94303 — email: MUS001FOS@myCFO.com\n>\n> **3. Account Advisor Information**\n> - Account advisors will serve as the owners/trustees in Section 2 (checked).\n> - Account Advisor B: **Ronald F. Gong**, Financial Advisor (myCFO), 2200 Geng Road, Suite 100, Palo Alto, CA — email irgong@mycfo.com & MUS001FOS@mycfo.com. Signed 07-08-2014.\n>\n> **4. Contribution Information**\n> - Contribution Type: marketable securities held outside of Vanguard Brokerage Services and mutual funds held outside Vanguard. Security/Fund Name: **TBD**.\n>\n> **5. Investment Option Recommendation**\n> - Multi-Fund Investment Options: Gift Preservation 50% / Moderate Growth 50%.\n>\n> **6. Succession Plan**\n> - Option 1 of 5: Keep the Account. 100%.\n> - Successor-Advisor A: **Antonio Gracias** (relationship to current account advisor: \"Friend\"); 200 South Michigan Avenue, Suite 1020, Chicago, IL 60604.\n>\n> **7. Related Donors:** [blank]\n>\n> **8. Signatures**\n> - Owner/Trustee A signature: Elon Musk, dated **07-09-2014**.\n> - \"I acknowledge that the gift of the property described in Section 4 will be **irrevocable and unconditional** when received and accepted by Vanguard Charitable. I acknowledge that I have read and agree to abide by the terms and conditions set forth in Vanguard Charitable's *Policies and guidelines* booklet…\"\n>\n> **9. Referral Information:** Financial Advisor.\n>\n> **10. Return Information:** Mail to Vanguard Charitable, P.O. Box 55766, Boston, MA 02205-5766; email donorservice@vanguardcharitable.org; fax 866-485-9414.\n\n## Commentary\n\nDX 1284 is the structural underlay of one of the most important defense soft-spots — and one of plaintiffs' standing problems. The same Vanguard Charitable DAF created by this agreement is the legal source from which the bulk of Musk's roughly $38M of OpenAI donations flowed: as Birchall conceded on Day 4 cross, \"**For every single one of those donations, it's somewhere in the chain, Mr. Musk was entitled to take a tax deduction**\" (`043026TT.txt:5413`). The \"irrevocable and unconditional\" language in Section 8 is the exact legal hook for plaintiffs' donor-advised-funds wrinkle: once the property entered the DAF, **Musk could not take it back**, only \"recommend\" grants — which goes directly to whether Musk has personal standing to seek \"restitution\" for charitable funds that ceased to be his in 2014. The form also names **Antonio Gracias** as successor-advisor (Musk's longtime board ally, then on Tesla's board) and **Ronald F. Gong / myCFO** as account advisor, prefiguring the same Birchall-era \"I'm not familiar with all of the legalities of donor advised funds\" answer (`043026TT.txt:5007`) that the defense exploited on Day 4. See also [[Key Themes]] § \"donor-advised funds / tax-deduction wrinkle.\"\n\n---\n\n*See also:* [[Day 4|Day 4 digest]] · [[Jared Birchall]] · [[Key Themes]] · [[Statute of Limitations]] · [[Standing]]\n"}
{"exhibit_id": "DX-1285", "exhibit": "DX 1285", "party": "Defense", "type": "Donor-advised-fund policies booklet (compilation)", "admitted_trial_day": "Day 4 (April 30, 2026)", "uploaded_box_pt": "2026-04-30T15:13:26", "uploader": "Morrison Foerster", "pages": 149, "size_bytes": 8926098, "source_pdf": "DX-1285.pdf", "pdf_url": "https://media.mts-in.com/DX-1285.pdf", "body_markdown": "# DX 1285 — Vanguard Charitable Policies and Guidelines (2014–2017)\n\n> A 149-page compilation of four annual editions of the **Vanguard Charitable Policies and Guidelines** booklet (versions effective in 2014, 2015, 2016, and 2017) — defense's documentary foundation for arguing that grants Musk made through his Vanguard Charitable donor-advised fund were governed by Vanguard's standard DAF rules, not by an OpenAI charitable-solicitation relationship.\n\n## Document type\n**Compilation document, 149 pages, ~8.5 MB.** Four sequential editions of Vanguard Charitable Endowment Program's \"Policies and guidelines\" booklet stacked in one file — the 2014 edition (Bates VC000339–375), the 2015 edition (VC000001–036, beginning with the 2015 \"10/1/2015\" investment date stamps and the explicit Wisconsin disclosure), the 2016 edition (VC000037–072, \"4/29/2016\" date stamps), and the 2017 edition (VC000038–072, \"10/1/2017\" date stamps, including the new \"Account abandonment\" structure and the \"Philanthropic Impact Fund\" rename). Marked \"CONFIDENTIAL\" with sequential VC000xxx Bates. Document ID DX-1285.\n\n## Logistics\n- **Trial admission:** Day 4 (April 30, 2026) — used by defense during Birchall direct/cross to establish the legal/operational framework governing Musk's Vanguard Charitable DAF grants to OpenAI. No prior wiki reference; admission day inferred from Box upload date.\n- **Box upload:** 2026-04-30 15:13:26 PT — Day 4 mid-afternoon batch (clustered with DX 539, DX 646, DX 1156).\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~8.5 MB, 149 pages — the largest DX in this chunk.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `1285.pdf`.\n\n## Transcribed text\n\n> [Document is 149 pages of standard donor-advised-fund (DAF) policies and is too long for a verbatim transcription within the page limit. Substantive provisions excerpted below; remainder is repetitive across the four editions.]\n>\n> **About Vanguard Charitable** (cover language, materially identical across editions):\n>\n> > Vanguard Charitable Endowment Program (\"Vanguard Charitable\") is a nonprofit organization that fulfills its mission to increase philanthropy [later editions: \"and maximize its impact over time\"] by administering a donor-advised fund — a tax-effective way to consolidate, accrue, and grant assets to charity. Vanguard Charitable was incorporated in the Commonwealth of Pennsylvania and is recognized by the IRS as a tax-exempt public charity under sections 501(c)(3), 509(a)(1), and 170(b)(1)(A)(vi) of the Internal Revenue Code. Vanguard Charitable's primary activity is to sponsor donor-advised accounts (\"philanthropic accounts\"). As provided in the Internal Revenue Code's §4966(d)(2), a donor-advised account is: (1) a fund or account owned and controlled by a sponsoring organization, (2) which is separately identified by reference to contributions of the donor or donors, and (3) where the donor (or a person appointed or designated by the donor) has or reasonably expects to have advisory privileges over the distribution or investments of the assets. Vanguard Charitable was founded by The Vanguard Group, Inc. (\"Vanguard\"), as an independent, nonprofit, public charity… A majority of Vanguard Charitable's trustees are independent of Vanguard.\n>\n> **Contributions — Overview** (materially identical across editions):\n>\n> > Once accepted by Vanguard Charitable, contributions are unconditional and irrevocable. Contributions and any future related earnings are no longer the property of the donor; they are owned by Vanguard Charitable and will be used solely for charitable purposes. For this reason, when contributing, the donor cannot impose any restrictions or conditions that prevent Vanguard Charitable from freely and effectively using the gift to further its mission.\n>\n> **Account access and roles — Donor:**\n>\n> > Donor: Prior owner of assets contributed to Vanguard Charitable to fund an account… Donors who make a contribution to open an account receive the privilege to name the account and a maximum of two account advisors, recommend an initial investment allocation, and nominate successor-advisors. Additional donors, such as individuals or organizations, may contribute to the account once it is established.\n>\n> **Account access and roles — Account advisor:**\n>\n> > Account advisor: Individual with ongoing account privileges and access, primary responsibility for certain account activities, and contact with Vanguard Charitable. Each account may have a maximum of two account advisors… Account advisors receive equal access and privileges to their account. While some account activities must be initiated and approved by all account advisors, many may be completed by one account advisor.\n>\n> **Grant recommendations — Prohibited grants** (later editions):\n>\n> > Grants from philanthropic accounts must be made exclusively for charitable purposes. Grants cannot result in more than incidental benefit for a donor, account advisor, interested party, or family members of the aforementioned. Prohibited benefits include, but are not limited to the: Payment of tuition. Fulfillment of legally binding pledges or other obligation. Receipt of anything of value in return for the grants, such as free dinner or goods or services… Under rules enacted by the Pension Protection Act of 2006, the IRS imposes stiff penalties on any donor, account advisor, or interested party who recommends a grant that results in impermissible benefit or receives impermissible benefit. Vanguard Charitable will take other action as appropriate, including but not limited to: Requiring the recipient nonprofit organization to repay the grant funds to Vanguard Charitable. Terminating philanthropic account privileges. Transferring the account balance to the General Fund [later: The Philanthropic Impact Fund].\n>\n> **Grant recommendations — Pledges and other legal obligations** (2015+ editions):\n>\n> > Generally, grants from philanthropic accounts cannot be used to pay any legal obligation that a Disqualified Person owes to a charity. An existing grant agreement or legally binding pledge to make a gift between a donor and charity are examples of such an obligation. A pledge is a promise that binds an individual to make a charitable contribution, and satisfying a pledge through a grant from a Vanguard Charitable account constitutes impermissible benefit to that individual. In addition, donors cannot commit to give assets that they legally do not own. As an alternative, donors may express \"intent to recommend\" and make nonbinding grant recommendations from their accounts. A donor's \"intent\" cannot imply that a grant is guaranteed to be issued from Vanguard Charitable.\n>\n> **Grant recommendations — Impermissible benefit:**\n>\n> > Grants from Vanguard Charitable are for the public good and cannot result in more than incidental benefit accruing to any specific individual, nor can they be restricted for a specific individual. For this reason, grants will not be approved if they will result in any benefit to the donor that would have the effect of reducing the charitable deduction if the donor had made the contribution directly from personal funds.\n>\n> **Tax considerations — Deductibility:**\n>\n> > Donors may qualify for a tax deduction for their contributions to Vanguard Charitable. Donors who itemize tax deductions are eligible for the maximum deduction allowed by law for charitable contributions. Only the owner of the donated asset is eligible for a tax deduction; account advisors may not claim a contribution into their accounts made by other individuals or organizations. Grants from Vanguard Charitable and account-balance fluctuations due to market changes are not tax-deductible.\n>\n> **Legacy options — Crafting a giving legacy:**\n>\n> > Account advisors may establish their giving legacies by recommending succession plans — plans which are enacted when advisors are deceased or unable or unwilling to manage account processes. In all instances, Vanguard Charitable remains the sole owner of the philanthropic account and has full discretion over the account and its investments and grants.\n>\n> [Edition-specific differences: the 2017 edition introduces explicit \"Account abandonment\" rules and replaces references to the \"General Fund\" with \"The Philanthropic Impact Fund.\" Maintenance fee changes from $100 (2014 edition) to $250 (2015+ editions). Mailing address changes from Boston, MA (2014–2016) to Warwick, RI (2017).]\n\n## Commentary\n\nDX 1285 is the largest exhibit in the chunk by file size (~8.5 MB) and one of the most quietly load-bearing for the defense. Plaintiffs' [[PX 103]] § 17510.6 charitable-solicitation theory depends on characterizing 2017–2020 Musk-side payments to OpenAI (including the **\"$290k monthly Fidelity grants\"** and the rent reimbursement Clark proposed) as *charitable solicitations* of the Musk Foundation/Birchall by OpenAI. Defense uses DX 1285 to make a structural counter-argument: many of the Musk-side payments to OpenAI flowed not from Musk personally or from the Musk Foundation but from **donor-advised funds** — Vanguard Charitable and Fidelity Charitable — and those DAFs are themselves the legal \"donors\" once contributions are made. Per the booklet, **\"Once accepted by Vanguard Charitable, contributions are unconditional and irrevocable. Contributions and any future related earnings are no longer the property of the donor; they are owned by Vanguard Charitable and will be used solely for charitable purposes.\"** Defense uses this language to argue: (i) the funds OpenAI received were not solicited from Musk in a § 17510.6 sense but were grants from a public charity (Vanguard Charitable) that owned the assets outright; (ii) Musk could not \"earmark\" or condition grants without violating the booklet's \"**impermissible benefit**\" and \"**pledges and other legal obligations**\" provisions, undermining plaintiffs' theory of Musk-controlled charitable conditions; and (iii) the § 4966(d)(2) IRC framework recited verbatim in the booklet preempts state-law solicitation theories that would treat the donor as the source of solicitation. The 2014 edition was selected as the starting year because it is the edition in force at the time Musk's largest documented grants began flowing to OpenAI nonprofit (via Vanguard Charitable and Fidelity Charitable). Cross-reference: [[PX 103]] (July 2020 Clark→Birchall reimbursement email) and [[PX 87]] (June 2017 YC.org tax-acknowledgment letter on the $5M).\n\n---\n*See also:* [[Day 4|Day 4 digest]] · [[PX 103]] · [[PX 87]] · [[Statute of Limitations]] · [[Pioneer Building]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-1290", "exhibit": "DX 1290", "party": "Defense", "type": "Text messages", "admitted_trial_day": "Day 8 (May 6, 2026)", "uploaded_box_pt": "2026-05-06T14:57:02", "uploader": "Morrison Foerster", "pages": 2, "size_bytes": 88345, "source_pdf": "DX-1290.pdf", "pdf_url": "https://media.mts-in.com/DX-1290.pdf", "body_markdown": "# DX 1290 — Nov 17, 2023 Zilis → Altman supportive text on the day of his firing\n\n> Shivon Zilis's brief 6:21 PM (UTC) text to Sam Altman on November 17, 2023 — the day the OpenAI nonprofit board fired him — saying \"**I just wanted to say I hope you are ok. I have no idea what's going on but you've been an awesome person in literally every interaction we've had and I care about you as a person first and foremost.**\"\n\n## Document type\n**Text messages, Cellebrite-style \"Short Message Report\" extraction.** Two pages: cover sheet (1 conversation, 2 messages, 3 participants — Sam Altman, Shivon Zilis, and a System Message; handles redacted) and page 2 transcript with timestamps in GMT+00:00. Conversation ID `CHAT - 96518_003.05.0000001 - 00491 - 2023/11/17`. Bates OPENAI_MUSK00018578–18579. Document ID DX-1290.\n\n## Logistics\n- **Trial admission:** Day 8 (May 6, 2026). Pre-trial wiki reference: **\"DX 1290 — Nov 17, 2023 Zilis-Altman supportive texts.\"** Used during the Zilis live direct.\n- **Box upload:** 2026-05-06 14:57:02 PT — Day 8 mid-afternoon defense batch (clustered with [[DX 627]], [[DX 651]], [[DX 712]], [[DX 1017]]).\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~86 KB, 2 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `1290.pdf`.\n\n## Transcribed text\n\n> **Conversation ID:** CHAT - 96518_003.05.0000001 - 00491 - 2023/11/17\n> **Date Range:** 11/17/2023\n> **Participants:** sam altman [redacted] · shivon zilis [redacted] · System Message\n>\n> *(Times shown in GMT +00:00.)*\n>\n> **SZ** (shivon zilis) — 11/17/2023, 6:21 PM\n> Obviously no need to reply but I just wanted to say I hope you are ok. I have no idea what's going on but you've been an awesome person in literally every interaction we've had and I care about you as a person first and foremost. Sending all of my positive vibes your way.\n>\n> **SM** (System Message) — 11/17/2023, 6:21 PM\n> shivon zilis disabled disappearing messages.\n\n## Commentary\n\nDX 1290 is a quiet defense exhibit with two interlocking purposes. First, the **substance**: on the day of Sam Altman's firing — when Musk publicly suggested Altman had been hiding a dangerous AGI from the board (see [[Key Themes]]) — Musk's longtime collaborator and the mother of his children was privately texting Altman: \"**you've been an awesome person in literally every interaction we've had and I care about you as a person first and foremost.**\" Defense uses this to undercut the depiction of Altman as a manipulative bad actor and to show that even those closest to Musk's circle did not adopt his post-2022 framing. Second, the **System Message**: \"*shivon zilis disabled disappearing messages*\" — Zilis affirmatively *preserved* this exchange. Defense reads that as evidence she was aware these messages were potentially discoverable and chose to keep the record visible. The exhibit pairs naturally with [[DX 1017]] (Zilis's Feb. 2023 \"**father of your babies starts a competitive effort**\" resignation text) — the same speaker, nine months later, comforting the same target Musk would soon be litigating against. Cross-reference: see [[Day 8|Day 8 digest]] for Toner deposition video on the November 2023 board events; PX 301 (Murati Slack) for adjacent governance context.\n\n---\n*See also:* [[Day 8|Day 8 digest]] · [[Shivon Zilis]] · [[Sam Altman]] · [[DX 1017]] · [[PX 301]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-1291", "exhibit": "DX 1291", "party": "Defendants", "type": "Text messages", "admitted_trial_day": "Day 8 (May 6, 2026)", "uploaded_box_pt": "2026-05-06T14:57:09", "uploader": "Morrison Foerster", "pages": 7, "size_bytes": 1572864, "source_pdf": "DX-1291.pdf", "pdf_url": "https://media.mts-in.com/DX-1291.pdf", "body_markdown": "# DX 1291 — July 12, 2017 Brockman-Zilis chat: \"Btw do you think they'd go for reverse merger?\" / \"We just need him to want more than a billion dollar exit\"\n\n> A 94-message single-day chat thread (G2BBBW chat group, four participants) between Greg Brockman and Shivon Zilis on July 12, 2017 — six weeks before Brockman's \"**this is the only chance we have to get out from under Elon … take me to $1,000,000,000**\" PX 151 journal entry, and ten weeks before Musk's PX 157 \"final straw\" email. Brockman floats a \"**reverse merger**\" structure to absorb [[Cerebras]]; Zilis says it's \"**a good thing to run by E**\" because \"**these guys are highly likely to be morally swayed for good. And could save over a billion dollars if done right**.\" Both then dissect what it would take to motivate the founder (\"**We just need him to want more than a billion dollar exit / Like more in life**\" — Zilis to Brockman re: the Cerebras founder).\n\n## Document type\n**Text messages.** A \"Short Message Report\" extract: 1 conversation (chat name \"G2BBBW\"), 94 messages, 4 participants, all dated 7/12/2017. Identified participants: Greg Brockman (chat owner / \"Me\"), Shivon Zilis (SZ), and two redacted participants. Bates OPENAI_MUSK00022838–22844.\n\n## Logistics\n- **Trial admission:** Day 8 (May 6, 2026) — Murati live; Zilis live; Toner video begins. Per the chunk wiki note: \"**DX 1291 — July 2017 Zilis-Brockman texts re: reverse merger**.\" Used by defense in connection with Zilis's live testimony to surface the **Cerebras reverse-merger** scenario as a real, contemporaneously-discussed July 2017 alternative to the for-profit conversion that ultimately took shape — framed by both sides as something \"to run by E[lon]\" before any commitment was made. See [[Key Themes]] § \"Greg Brockman's contributions\" for [[Cerebras]] and the parallel July 2017 Brockman conflict-of-interest theory.\n- **Box upload:** 2026-05-06 14:57:09 PT — Day 8 mid-afternoon batch.\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~1.5 MB, 7 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `1291.pdf`.\n\n## Transcribed text\n\n> **Short Message Report**\n> Conversations: 1 · Total Messages: **94** · Participants: 4 · Date Range: 7/12/2017\n>\n> Chat: **G2BBBW** — 94 messages on 7/12/2017\n> Participants: Greg Brockman (owner), Shivon Zilis, [+ two redacted]\n>\n> ---\n>\n> **Greg Brockman (\"Me\")** — 7/12/2017, 12:01 AM\n> Great, totally agreed\n>\n> **Greg Brockman** — 12:02 AM\n> All about inception\n>\n> **Greg Brockman** — 12:02 AM\n> I was really confused why Elon was asking me to explain\n>\n> **Greg Brockman** — 12:03 AM\n> (Also what was your specific realization?)\n>\n> **Shivon Zilis** — 12:03 AM\n> Because he wants to know. I mentioned after you left that there were more concrete structures to potentially discuss with you and he said for you to call him sometime soon\n>\n> **Shivon Zilis** — 12:03 AM\n> Oh. The amoral thing. They seemed embarrassed by it\n>\n> **Shivon Zilis** — 12:04 AM\n> Why did you start company?\n>\n> **Greg Brockman** — 12:04 AM\n> Ah!\n>\n> **Shivon Zilis** — 12:04 AM\n> I missed leading a team\n>\n> **Shivon Zilis** — 12:04 AM\n> But now oppty to be increasingly moral\n>\n> **Shivon Zilis** — 12:04 AM\n> **E is good at making people do that**\n>\n> **Shivon Zilis** — 12:04 AM\n> Sorry, leading them to it\n>\n> **Greg Brockman** — 12:04 AM\n> Yeah and had poor answer for what they want their impact to be\n>\n> **Greg Brockman** — 12:04 AM\n> \"Change the world!\"\n>\n> **Greg Brockman** — 12:04 AM\n> \"… to what?\"\n>\n> **Greg Brockman** — 12:04 AM\n> \"…\"\n>\n> **Shivon Zilis** — 12:06 AM\n> Ummmmm how about you do that\n>\n> **Shivon Zilis** — 12:06 AM\n> I will just make a chip?\n>\n> **Greg Brockman** — 12:09 AM\n> Heh. Btw what do you think are the prospects of a longer meeting with ilya/me/you/e soon?\n>\n> **Shivon Zilis** — 12:09 AM\n> If you want it we should make it happen\n>\n> **Shivon Zilis** — 12:09 AM\n> Longer meaning?\n>\n> **Shivon Zilis** — 12:10 AM\n> He is a time triage kinda guy\n>\n> **Shivon Zilis** — 12:10 AM\n> So if there is a thing to talk about he wants to make time\n>\n> **Shivon Zilis** — 12:10 AM\n> Just not great at the weekly meeting thing\n>\n> **Greg Brockman** — 12:12 AM\n> I think dinner next week could be pretty optimal\n>\n> **Greg Brockman** — 12:12 AM\n> Or an hour long meeting would work too\n>\n> **Greg Brockman** — 12:13 AM\n> But I think dinner is the right format\n>\n> **Shivon Zilis** — 12:15 AM\n> Hm. Next two weeks may be tight on dinner because he has an atypical schedule. I will ask Emma tomorrow\n>\n> **Shivon Zilis** — 12:15 AM\n> Prefer dinner maybe a bit later or one hour meetings soon if given choice?\n>\n> **Greg Brockman** — 12:15 AM\n> Later dinner\n>\n> **Shivon Zilis** — 12:15 AM\n> Gotcha\n>\n> **Greg Brockman** — 12:15 AM\n> More casual atmosphere is really helpful\n>\n> **Shivon Zilis** — 12:16 AM\n> Isn't it?\n>\n> **Shivon Zilis** — 12:16 AM\n> It's hard with him because he gets in game time mode and it's all tactical. He was so patient tonight\n>\n> **Greg Brockman** — 12:16 AM\n> Yeah was actually surprised he wasn't talking more\n>\n> **Shivon Zilis** — 12:17 AM\n> That was so nuts. When he just let it unfold\n>\n> **Greg Brockman** — 12:18 AM\n> Interesting, why nuts?\n>\n> **Shivon Zilis** — 12:18 AM\n> You should def call him though\n>\n> **Shivon Zilis** — 12:18 AM\n> **Meaning few people know when to be silent so information reveals itself**\n>\n> **Shivon Zilis** — 12:18 AM\n> He was silent, asked the right question, then was silent\n>\n> **Greg Brockman** — 12:18 AM\n> Ah I see\n>\n> **Greg Brockman** — 12:18 AM\n> Very good point\n>\n> **Greg Brockman** — 12:21 AM\n> Anyway super glad to have you involved in all these things. Really useful to have a tighter feedback loop to E and feel like I have much better perspective and thoughts as a result of our convos.\n>\n> **Shivon Zilis** — 12:25 AM\n> Oh that's so nice to hear. The fact that I can be helpful makes me so happy. You know how much I care\n>\n> **Shivon Zilis** — 12:28 AM\n> Oh! Only tiny piece of feedback, just linguistic\n>\n> **Shivon Zilis** — 12:29 AM\n> For Andrew, would be useful to make sure we use language around \"**AGI done for the best possible future**\" vs \"**AGI**\" in and of itself\n>\n> **Shivon Zilis** — 12:30 AM\n> Implicit — but good to just weave that in cuz **DeepMind is building AGI too**\n>\n> **Greg Brockman** — 12:30 AM\n> Good point\n>\n> **Shivon Zilis** — 12:30 AM\n> Does that make sense? Hard over text\n>\n> **Greg Brockman** — 12:30 AM\n> Yes\n>\n> **Greg Brockman** — 12:34 AM\n> **Btw do you think they'd go for reverse merger?**\n>\n> **Shivon Zilis** — 12:40 AM\n> I think not yet but I think it's a good thing to run by E. Reason being, he'll either be into it or not into it, and if he's into it, **I think these guys are highly likely to be morally swayed for good. And could save over a billion dollars if done right** / it may have not otherwise been done because of price point\n>\n> **Greg Brockman** — 12:41 AM\n> cool\n>\n> **Shivon Zilis** — 12:41 AM\n> What do you think?\n>\n> **Greg Brockman** — 12:44 AM\n> i think you're right, but also there's a second reason they might want to do it: **we can make their chip way more valuable by building to it** (btw if no exclusivity seems kinda crazy for us to invest a bunch in making it better…). So even in non-crazy world, still can be very valuable to their existing product to have it. Obviously lots of questions about structure and motivation to be sorted out.\n>\n> **Shivon Zilis** — 12:45 AM\n> Yep I hear you\n>\n> **Shivon Zilis** — 12:45 AM\n> Just heard that **last round has post of 700**\n>\n> **Shivon Zilis** — 12:45 AM\n> One that is closing now\n>\n> **Greg Brockman** — 12:45 AM\n> yep\n>\n> **Greg Brockman** — 12:46 AM\n> (I'm an investor!)\n>\n> **Shivon Zilis** — 12:46 AM\n> Oh nice\n>\n> **Shivon Zilis** — 12:46 AM\n> Who is bulk of that money?\n>\n> **Greg Brockman** — 12:46 AM\n> Also if I were them, I'd wonder how to stay ahead\n>\n> **Shivon Zilis** — 12:46 AM\n> He told E more before, which was interesting\n>\n> **Greg Brockman** — 12:46 AM\n> Since if they succeed everyone will copy them\n>\n> **Greg Brockman** — 12:46 AM\n> And also like… how big is their market rally?\n>\n> **Greg Brockman** — 12:46 AM\n> *really\n>\n> **Shivon Zilis** — 12:47 AM\n> **We just need him to want more than a billion dollar exit**\n>\n> **Greg Brockman** — 12:47 AM\n> Yeah\n>\n> **Shivon Zilis** — 12:47 AM\n> **Like more in life**\n>\n> **Greg Brockman** — 12:47 AM\n> yep\n>\n> **Greg Brockman** — 12:47 AM\n> He's already had the one exit\n>\n> **Shivon Zilis** — 12:47 AM\n> You're a boy, you know how male egos work better than I do\n>\n> **Shivon Zilis** — 12:47 AM\n> Yes that is in our favor definitely\n>\n> **Shivon Zilis** — 12:48 AM\n> Otherwise kinda no hope. You gotta get that done to have crazy pure motivations it seems\n>\n> **Shivon Zilis** — 12:48 AM\n> Btw male ego thing meant with a happy winky face haha\n>\n> **Greg Brockman** — 12:49 AM\n> most coming from… **Vy Capital Management Company Limited (dubai), tencent, and a bit from coatue (probably pro-rata)**\n>\n> **Greg Brockman** — 12:49 AM\n> So randos\n>\n> **Shivon Zilis** — 12:53 AM\n> Surprised SoftBank didn't jump\n>\n> **Shivon Zilis** — 2:12 PM\n> A few things for when you have a moment\n>\n> **Greg Brockman** — 2:15 PM\n> On a call but free in 15 mins\n>\n> **Shivon Zilis** — 2:16 PM\n> Not urgent\n>\n> **Shivon Zilis** — 2:16 PM\n> Let's touch based in next day or two tho\n>\n> **Shivon Zilis** — 2:41 PM\n> **Btw Demis cancelled his trip?**\n>\n> **Greg Brockman** — 2:42 PM\n> Apparently ya\n>\n> **Greg Brockman** — 3:04 PM\n> Free now btw\n>\n> **Shivon Zilis** — 3:41 PM\n> Stuck in meetings now\n>\n> **Shivon Zilis** — 3:42 PM\n> Will call you later\n>\n> **Greg Brockman** — 3:42 PM\n> Sg\n>\n> **Shivon Zilis** — 3:43 PM\n> Just wanted to catch you before you or Ilya pitch him anything\n>\n> **Shivon Zilis** — 3:43 PM\n> So you have context\n>\n> **Greg Brockman** — 3:45 PM\n> great. nothing calendared yet.\n>\n> **Shivon Zilis** — 11:40 PM\n> **Just getting home from Tesla**. Sorry I missed you earlier\n>\n> **Shivon Zilis** — 11:41 PM\n> Can chat now or tomorrow. Not urgent given next meeting seems to be next week but will fill you in when you have time\n\n[Exhibit reproduces the substantive portion of the 94-message July 12, 2017 chat. The redacted second-participant identifier next to Zilis would correspond to her personal phone number.]\n\n## Commentary\n\nDX 1291 is, in three respects, the densest single-day window into the Brockman-Zilis information channel into Musk in the summer of 2017. **Defense's read:** the chat shows Brockman and Zilis (i) **planning a \"longer meeting\" with Musk and Ilya**, (ii) **strategizing how to \"morally sway\" the [Cerebras] founder** — because Brockman thinks a **reverse merger** of OpenAI into Cerebras (or some structural integration) \"could save over a billion dollars if done right\" and \"make their chip way more valuable by building to it,\" and (iii) **explicitly calibrating language** around \"**AGI done for the best possible future**\" versus \"AGI\" in itself, \"cuz DeepMind is building AGI too.\" None of this is consistent with a stable nonprofit research organization preserving its charitable mission; all of it is consistent with the founders themselves, *six weeks before the August 2017 control negotiation*, looking for the right wrapper to (a) absorb a chip company and (b) sell the resulting story to Musk. **Plaintiffs' read:** the chat *is* the conflict of interest. Brockman is **a Cerebras investor** (\"I'm an investor!\" — 12:46 AM), and he is here strategizing with Musk's intermediary on how to engineer a transaction in which OpenAI either acquires or reverse-merges with the chip company in which he holds equity, on terms designed to make Cerebras's product \"way more valuable\" — the textbook self-dealing pattern (see [[Key Themes]] § \"Greg Brockman's contributions\" and the [[Brockman Journal]] entries on Cerebras). The \"**we just need him to want more than a billion dollar exit / Like more in life**\" line is plaintiffs' Exhibit A on what Brockman and Zilis actually thought of the [[Cerebras]] founder — that he could be \"morally swayed\" by aspirational language and a billion-plus payout, \"to have crazy pure motivations.\" The same dynamic, plaintiffs argue, is then redeployed against Musk himself in [[DX 631]] (Founder Series Tesla gifts), [[DX 653]] (Sutskever's \"Will a model 3 make you be willing to accept massively unfavorable terms?\"), and the August–September 2017 control sequence in [[PX 151]] / [[PX 156]] / [[PX 157]]. The \"**Demis cancelled his trip**\" reference is to **Demis Hassabis**, DeepMind founder, and confirms the parallel-track competitive frame Zilis deployed minutes earlier (\"DeepMind is building AGI too\"). The \"Vy Capital (dubai), tencent, … coatue\" line identifies the participants in Cerebras's then-current fundraising round at a \"post of 700\" (post-money $700M) valuation. The \"**Just getting home from Tesla**\" sign-off at 11:40 PM situates Zilis's parallel role inside Musk's portfolio companies — she was already operating across the Musk universe (Neuralink + Tesla + OpenAI) by July 2017.\n\n---\n*See also:* [[Day 8|Day 8 digest]] · [[DX 631]] · [[DX 653]] · [[DX 715]] · [[PX 151]] · [[PX 156]] · [[PX 157]] · [[Greg Brockman]] · [[Shivon Zilis]] · [[Cerebras]] · [[Brockman Journal]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-1293", "exhibit": "DX 1293", "party": "Defense (OpenAI/MS)", "type": "Stipulation", "admitted_trial_day": "Day 3 (April 29, 2026)", "uploaded_box_pt": "2026-04-29T15:37:12", "uploader": "OpenAI startup-fund defense firm", "pages": 11, "size_bytes": 527220, "source_pdf": "DX-1293.pdf", "pdf_url": "https://media.mts-in.com/DX-1293.pdf", "body_markdown": "# DX 1293 — Joint stipulation of undisputed facts (Musk v. Altman)\n\n> The parties' 56-paragraph joint stipulation of undisputed facts — the master skeleton of OpenAI's corporate history, the contribution ledger, the board roster, the Microsoft rounds, and the recapitalization timeline that frames every other exhibit in the case.\n\n## Document type\n**Stipulation.** Caption-less filing titled \"*Musk, et al.* v. *Altman, et al.* — Undisputed Facts.\" 56 numbered paragraphs across 11 pages. Includes (i) the OpenAI-entity family tree, (ii) the full board-service roster from 1/3/2016 through 1/14/2025, (iii) the line-by-line ledger of Musk's $37.8M–$38.2M in contributions through Vanguard DAF, Fidelity DAF, Musk Foundation, and YC Org, (iv) the four Tesla in-kind contributions, (v) the Microsoft 2019/2021/2023 capital-commitment / target-redemption math, and (vi) the recapitalization-into-PBC milestones through GPT-5.2.\n\n## Logistics\n- **Trial admission:** Day 3 (April 29, 2026) — Musk direct concludes; cross begins. Used as a foundational reference document during Musk's cross.\n- **Box upload:** 2026-04-29 15:37:12 PT — same-day upload.\n- **Uploader:** OpenAI startup-fund defense firm.\n- **File size:** ~515 KB, 11 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `1293 .pdf`.\n\n## Transcribed text\n\n> ***Musk, et al.* v. *Altman, et al.* — Undisputed Facts**\n>\n> 1. Plaintiff Elon Musk is an individual and a resident of Texas.\n> 2. Defendant Sam Altman is an individual and a resident of California.\n> 3. Defendant Greg Brockman is an individual and a resident of California.\n> 4. Defendant OpenAI, Inc., now known as The OpenAI Foundation, is a nonprofit corporation incorporated under the laws of Delaware, with its headquarters in California. OpenAI, Inc. is registered as an out-of-state nonprofit corporation with the California Secretary of State. OpenAI, Inc. is registered as a tax-exempt nonprofit organization under federal and California tax laws.\n> 5. Defendant OpenAI, L.P. was a limited partnership formed under the laws of Delaware on September 19, 2018 and was registered as an out-of-state limited partnership with the California Secretary of State, with its headquarters in California. OpenAI, L.P. was an affiliate of OpenAI, Inc.\n> 6. OpenAI Group PBC is a public benefit corporation incorporated under the laws of Delaware and formed in connection with the recapitalization of the OpenAI for-profit enterprise on October 28, 2025.\n> 7. Defendant OpenAI OpCo, LLC is a limited liability company incorporated under the laws of Delaware on January 23, 2023. OpenAI OpCo, LLC succeeded OpenAI, L.P. and is an affiliate of OpenAI Group PBC.\n> 8. OpenAI Global, LLC is a limited liability company incorporated under the laws of Delaware on December 28, 2022. Investors' interests in OpenAI, L.P. were transferred to OpenAI Global, LLC on or around that same date.\n> 9. Defendants OpenAI, LLC, OpenAI GP, LLC, OAI Corporation, OpenAI Holdings LLC, OpenAI Startup Fund Management, LLC, OpenAI Startup Fund GP I, LLC, OpenAI Startup Fund I, LP, OpenAI Startup Fund SPV GP I, LLC, OpenAI Startup Fund SPV GP II, LLC, OpenAI Startup Fund SPV GP III, LLC, OpenAI Startup Fund SPV GP IV, LLC, OpenAI Startup Fund SPV I, LP, OpenAI Startup Fund SPV II, LP, OpenAI Startup Fund SPV III, LP, OpenAI Startup Fund SPV IV, LP, Aestas Management Company, LLC, and Aestas, LLC are or were limited liability companies, corporations, or limited partnerships formed and incorporated under the laws of Delaware.\n> 10. Defendant Microsoft Corporation is a corporation formed under the laws of Washington with its principal place of business in Washington.\n> 11. Elon Musk, Sam Altman, Greg Brockman, and Ilya Sutskever were each co-founders of OpenAI.\n> 12. OpenAI, Inc. was incorporated as a nonprofit, nonstock corporation in December 2015.\n> 13. OpenAI, Inc. filed its original certificate of incorporation with the Delaware Secretary of State on December 8, 2015.\n> 14. The following individuals have served on OpenAI, Inc.'s board:\n>\n> | Individual | Dates of Service on Board |\n> | --- | --- |\n> | Sam Altman | 1/3/2016 to 11/16/2023; 3/8/2024 to Present |\n> | Greg Brockman | 6/30/2017 to 11/16/2023 |\n> | Chris Clark | 1/3/2016 to 9/1/2017 |\n> | Adam D'Angelo | 9/21/2018 to Present |\n> | Sue Desmond-Hellmann | 3/8/2024 to Present |\n> | Reid Hoffman | 3/11/2019 to 2/6/2023 |\n> | Will Hurd | 4/30/2021 to 5/31/2023 |\n> | Holden Karnofsky | 6/30/2017 to 6/30/2021 |\n> | Zico Kolter | 8/8/2024 to Present |\n> | Tasha McCauley | 11/30/2018 to 11/29/2023 |\n> | Elon Musk | 1/3/2016 to 2/21/2018 |\n> | Paul Nakasone | 6/13/2024 to Present |\n> | Adebayo Ogunlesi | 1/14/2025 to Present |\n> | Patrick Scaglia | 1/3/2016 to 6/30/2017 |\n> | Nicole Seligman | 3/8/2024 to Present |\n> | Lawrence Summers | 11/29/2023 to 11/19/2025 |\n> | Fidji Simo | 3/8/2024 to 5/9/2025 |\n> | Ilya Sutskever | 6/30/2017 to 11/29/2023 |\n> | Bret Taylor | 11/29/2023 to Present |\n> | Helen Toner | 7/9/2021 to 11/29/2023 |\n> | Sue Yoon | 9/21/2018 to 9/30/2019 |\n> | Shivon Zilis | 1/10/2020 to 2/25/2023 |\n>\n> 15. From 2016 through September 14, 2020, Musk was the original source of between $37,799,400 (OpenAI Defendants' figure) and $38,191,066 (Plaintiff's figure) in cash and in-kind contributions to OpenAI, Inc. Of those amounts, (i) $10,500,000 of the cash contributions were paid by Elon Musk or the Musk Foundation to OpenAI's fiscal sponsor, YC.org (\"YC Org\"), and thereafter paid by YC Org to OpenAI, Inc.; (ii) $11,510,000 of the cash contributions were paid to OpenAI, Inc. following Musk's recommendations by the Vanguard Charitable Endowment Program (\"Vanguard DAF\") or the Fidelity Investments Charitable Gift Fund (\"Fidelity DAF\"); and (iii) $15,527,000 of the cash contributions were paid to YC Org following Musk's recommendations by Vanguard DAF, and thereafter paid to OpenAI, Inc. YC Org served as OpenAI, Inc.'s fiscal sponsor to receive tax-exempt charitable contributions before OpenAI, Inc.'s own tax-exempt status was approved. The in-kind contributions in the form of four Tesla vehicles were made directly by Elon Musk to OpenAI, Inc.\n> 16. The following chart lists cash contributions for which Musk was the original funding source and OpenAI, Inc., was the ultimate recipient:\n>\n> | Approx. Date | Payor | Payee | Amount |\n> | --- | --- | --- | --- |\n> | 5/27/2016 | Elon Musk | YC Org | $500,000.00 |\n> | 6/3/2016 | Vanguard DAF | YC Org | $5,000,000.00 |\n> | 8/22/2016 | Vanguard DAF | YC Org | $4,500,000.00 |\n> | 9/23/2016 | Vanguard DAF | YC Org | $142,000.00 |\n> | 10/18/2016 | Vanguard DAF | YC Org | $142,000.00 |\n> | 11/14/2016 | Vanguard DAF | YC Org | $750,000.00 |\n> | 11/16/2016 | Vanguard DAF | YC Org | $142,000.00 |\n> | 12/1/2016 | Vanguard DAF | YC Org | $4,250,000.00 |\n> | 12/16/2016 | Vanguard DAF | YC Org | $142,000.00 |\n> | 1/18/2017 | Vanguard DAF | YC Org | $142,000.00 |\n> | 2/16/2017 | Vanguard DAF | YC Org | $142,000.00 |\n> | 2/27/2017 | Musk Foundation | YC Org | $5,000,000.00 |\n> | 3/17/2017 | Vanguard DAF | YC Org | $175,000.00 |\n> | 4/19/2017 | Vanguard DAF | OpenAI, Inc. | $175,000.00 |\n> | 5/16/2017 | Vanguard DAF | OpenAI, Inc. | $175,000.00 |\n> | 5/26/2017 | Musk Foundation | YC Org | $5,000,000.00 |\n> | 6/16/2017 | Vanguard DAF | OpenAI, Inc. | $175,000.00 |\n> | 7/18/2017 | Fidelity DAF | OpenAI, Inc. | $175,000.00 |\n> | 8/14/2017 | Fidelity DAF | OpenAI, Inc. | $175,000.00 |\n> | 9/15/2017 | Fidelity DAF | OpenAI, Inc. | $175,000.00 |\n> | 9/29/2017 | Fidelity DAF | OpenAI, Inc. | $85,000.00 |\n> | 10/16/2017 | Fidelity DAF | OpenAI, Inc. | $235,000.00 |\n> | 11/14/2017 | Fidelity DAF | OpenAI, Inc. | $235,000.00 |\n> | 12/14/2017 | Fidelity DAF | OpenAI, Inc. | $235,000.00 |\n> | 1/18/2018 | Fidelity DAF | OpenAI, Inc. | $290,000.00 |\n> | 2/20/2018 | Fidelity DAF | OpenAI, Inc. | $390,000.00 |\n> | 3/14/2018 – 9/14/2020 | Fidelity DAF | OpenAI, Inc. | $290,000.00 monthly (with intervening months at $290,000.00 each) |\n>\n> *[Full month-by-month $290,000 Fidelity DAF schedule continues from 3/14/2018 through 9/14/2020 — every entry $290,000.00 except 2/20/2018 ($390,000) — totaling the stipulated $11.51M Vanguard/Fidelity DAF subset.]*\n>\n> 17. The following chart lists in-kind contributions from Musk to OpenAI, Inc., consisting of four Tesla vehicles and vehicle upgrades:\n>\n> | Approx. Date | Sender | Recipient | Amount |\n> | --- | --- | --- | --- |\n> | 10/2/2017 | Elon Musk | OpenAI, Inc. | $248,295 |\n> | 1/24/2018 | Elon Musk | OpenAI, Inc. | $14,105 |\n>\n> 18. From 2016 to 2020, OpenAI, Inc. received the following charitable contributions from donors other than Musk: (i) in 2016, $3,784,637 from Sam Altman; (ii) in 2017, $5 million from Aphorism Foundation, $100,000 from a donor's trust DAF, and $10 million from Good Ventures Foundation; (iii) in 2018, $500,000 from Alameda Research, $600,000 from Amazon Web Services (\"AWS\"), $5 million from Aphorism Foundation, $10,000 from BLTE, LLC, $50,000 from Fidelity Charitable, $20,008,279 from Gabe Newell, and $20 million from Good Ventures Foundation; and (iv) in 2019, $100,000 from AWS and $30 million from Silicon Valley Community Foundation. OpenAI, Inc. received $5 million from the Schwab Fund for Charitable Giving in 2023.\n> 19. In or around June 2016, Musk Industries LLC (\"Musk Industries\") and Bridgeton Pioneer Property LLC (\"Landlord\") executed a lease for the Pioneer Building for a period of 10 years. Around the same time, Musk Industries and OpenAI, Inc. entered into a tenancy at will agreement.\n> 20. As an artificial intelligence company, OpenAI, Inc. required computer processing resources, known as \"compute,\" to perform artificial intelligence training and research.\n> 21. In 2016, Microsoft and OpenAI, Inc. negotiated, executed, and subsequently announced publicly an agreement under which OpenAI, Inc. paid Microsoft a substantially discounted price for cloud computing services for artificial intelligence research and development.\n> 22. On August 11, 2017, OpenAI's technology defeated the world's top-ranked human player in a one-versus-one tournament of the Dota 2 video game.\n> 23. On August 28, 2017, OpenAI, Inc. filed an initial registration form with the California Attorney General's office to register as a foreign nonprofit corporation doing business in the state of California.\n> 24. On February 21, 2018, Musk resigned from OpenAI, Inc.'s board of directors.\n> 25. On April 9, 2018, OpenAI, Inc. posted a \"Charter\" on its website.\n> 26. OpenAI released GPT-1 on June 8, 2018.\n> 27. On September 19, 2018, OpenAI, L.P., originally called SummerSafe, L.P., was formed as a limited partnership under Delaware law. OpenAI, L.P. was initially governed by the Limited Partnership Agreement dated October 10, 2018. Investors in OpenAI, L.P. received limited partnership interests that entitled them each to receive a target redemption amount equal to a multiple of their investment. OpenAI, Inc. also received an uncapped residual equity interest in the partnership that entitled it to receive all profits after the limited partners' target redemption amounts were satisfied.\n> 28. On March 11, 2019, OpenAI publicly announced the launch of OpenAI, L.P.\n> 29. At or around the same time, OpenAI, Inc. transferred substantially all of its intellectual property to OpenAI, L.P., and most of the individuals who had been employed by OpenAI, Inc. became employees of OpenAI, L.P. OpenAI, Inc. received a limited partnership interest based on the appraised value of OpenAI, Inc.'s asset contribution to OpenAI, L.P. The target redemption amount applicable to OpenAI, Inc.'s asset contribution was ultimately set at $6,082,908,300, reflecting a target redemption multiple of 100 times the value of OpenAI, Inc.'s asset contribution.\n> 30. In its first fundraising round, which completed on March 1, 2019, OpenAI, L.P. received $133 million in capital commitments from outside investors, referred to as First Close Limited Partners (\"FCLP\"). The target redemption amount applicable to all FCLP capital commitments was $13.3 billion, reflecting a target redemption multiple of 100 times each FCLP's capital commitment.\n> 31. In April 2019, Altman became CEO of OpenAI.\n> 32. On July 2, 2019, Microsoft committed $1 billion in capital to OpenAI, L.P. The target redemption amount applicable to Microsoft's 2019 capital commitment was $20 billion, reflecting a target redemption multiple of 20 times Microsoft's capital commitment.\n> 33. On July 2, 2019, Microsoft, OpenAI, Inc. and OpenAI, L.P. executed a Joint Development and Collaboration Agreement (\"JDCA\") to govern their commercial partnership.\n> 34. OpenAI, Inc.'s board approved Microsoft's 2019 capital commitment and the 2019 JDCA.\n> 35. Prior to entering into a commercial relationship with OpenAI, Inc. and OpenAI, L.P., Microsoft conducted due diligence, including a review by Microsoft's outside counsel of documents provided by OpenAI related to OpenAI, Inc.'s and OpenAI, L.P.'s governance, structure, capitalization, and tax status.\n> 36. On July 22, 2019, both OpenAI and Microsoft issued public statements announcing their commercial partnership.\n> 37. OpenAI released the full GPT-2 model on November 5, 2019.\n> 38. On April 23, 2020, OpenAI, Inc. filed an amended certificate of incorporation with the Delaware Secretary of State.\n> 39. OpenAI released GPT-3 on May 28, 2020.\n> 40. On March 6, 2021, Microsoft agreed to provide an additional $2 billion worth of capital commitments to OpenAI, L.P., including credits for OpenAI's compute spend. Microsoft's target redemption amount for that capital commitment was $12 billion, reflecting a target redemption multiple of 6 times Microsoft's capital commitment. The target redemption multiple for all other limited partners remained 100 times their capital commitments.\n> 41. On March 5, 2021, Microsoft, OpenAI, Inc., and OpenAI, L.P. executed an Amended and Restated Joint Development and Collaboration Agreement.\n> 42. OpenAI, Inc.'s nonprofit board approved Microsoft's 2021 capital commitment and the 2021 JDCA, at the recommendation of a committee formed to review the transaction.\n> 43. Brockman became President of OpenAI on or around May 5, 2022.\n> 44. OpenAI released ChatGPT on November 30, 2022.\n> 45. On January 23, 2023, OpenAI's employee vehicle, Aestas, L.P. (now known as Aestas, LLC), became a member of OpenAI Global, LLC, and received a membership interest entitling it to receive a target redemption amount of $150 billion.\n> 46. On January 23, 2023, Microsoft agreed to provide an additional $10 billion worth of capital commitments to OpenAI Global, LLC, including credit for OpenAI's compute spend. Microsoft's target redemption amount for that investment was $60 billion, reflecting a target redemption multiple of 6 times Microsoft's capital commitment.\n> 47. On January 23, 2023, Microsoft, OpenAI, Inc., and OpenAI OpCo, LLC executed a Second Amended and Restated Joint Development and Collaboration Agreement to amend the terms of their commercial relationship.\n> 48. On or around January 20, 2023, OpenAI, Inc.'s nonprofit board unanimously approved Microsoft's 2023 capital commitment and the 2023 JDCA.\n> 49. OpenAI released GPT-4 on March 14, 2023.\n> 50. On April 10, 2023, OpenAI and Microsoft executed a Second Amended and Restated Limited Liability Company Agreement of OpenAI Global, LLC that further reflects Microsoft's 2023 capital commitment.\n> 51. On August 5, 2024, Plaintiff filed this lawsuit against several OpenAI entities, Altman, and Brockman.\n> 52. On November 14, 2024, Plaintiff filed an amended complaint adding, among other things, claims against Microsoft.\n> 53. OpenAI released GPT-4.5 on February 27, 2025.\n> 54. Plaintiff filed the operative second amended complaint on May 22, 2025.\n> 55. On October 28, 2025, OpenAI, Inc. recapitalized its for-profit subsidiaries into a public benefit corporation, OpenAI Group PBC, incorporated under Delaware law. Investors in the for-profit LLC entities received equity stakes in OpenAI Group PBC.\n> 56. OpenAI released GPT-5.2 on December 11, 2025.\n\n## Commentary\n\nDX 1293 is the case's quietest but most load-bearing exhibit: every other timeline-anchoring fact in the trial — Birchall's \"$38,191,066\" headline ([[Key Themes#The $1B / $38M gap]]), Brockman's 6/30/2017 board start, Musk's 2/21/2018 resignation, the SummerSafe-to-OpenAI-LP rename, the **20× Microsoft 2019** vs. **6× Microsoft 2023** target-redemption multiples that anchored Wu's Day 7 deposition ([[Key Themes#\"$250 billion\" — the target redemption stack]]), and the October 28, 2025 PBC recapitalization — flows from this stipulation. The DAF ledger in ¶16 is the documentary spine of the [[Key Themes#The \"donor-advised funds\" / tax-deduction wrinkle|\"donor-advised funds\" / tax-deduction wrinkle]]: most of the $38M flowed from Vanguard DAF and Fidelity DAF after Musk had already taken the deduction. Paragraph 19's confirmation that **Musk Industries LLC** — not OpenAI — held the Pioneer Building lease is the foundation for the [[Key Themes#The Pioneer Building|§17510.6 charitable-solicitation theory]]; ¶18's \"$3,784,637 from Sam Altman\" in 2016 is plaintiffs' counter to the \"**Altman contributed less than 10 percent**\" line. This is the document defense referred to whenever a witness was asked to confirm a date or dollar figure — and it is the structural map for every other DX/PX exhibit in the wiki.\n\n---\n*See also:* [[Day 3|Day 3 digest]] · [[Key Themes]] · [[Brockman Journal]] · [[Quotes]]\n"}
{"exhibit_id": "DX-1444", "exhibit": "DX 1444", "party": "Defense", "type": "Tweet", "admitted_trial_day": "Day 3 (April 29, 2026)", "uploaded_box_pt": "2026-04-29T15:37:43", "uploader": "Morrison Foerster", "pages": 1, "size_bytes": 376127, "source_pdf": "DX-1444.pdf", "pdf_url": "https://media.mts-in.com/DX-1444.pdf", "body_markdown": "# DX 1444 — March 4, 2026 Elon Musk tweet: \"Tesla will be one of the companies to make AGI and probably the first to make it in humanoid/atom-shaping form\"\n\n> Musk's **eight-week-pre-trial** tweet — March 4, 2026, six weeks before opening statements — staking Tesla's claim to be \"one of the companies to make AGI\" and \"probably the first to make it in humanoid/atom-shaping form.\" The single most-recent piece of \"parallel AGI play\" evidence in the case.\n\n## Document type\n**Tweet** — screenshot of an X (Twitter) post by @elonmusk, posted 1:15 AM, March 4, 2026, with a \"Follow\" button visible (indicating the screenshot was taken from a non-follower view). 57.3M views, 10K replies, 11K reposts, 132K likes, 6.3K bookmarks. Bates none visible; the bottom-right corner carries the standard DX-1444 Defendant exhibit stamp.\n\n## Logistics\n- **Trial admission:** Day 3 (April 29, 2026) — Musk direct concludes; cross begins. Per the wiki reference index: \"DX 1444 — Mar 2026 Musk tweet (Tesla will be one of the companies to make AGI).\" Used by defense in the [[Key Themes]] §\"Tesla / xAI as Musk's own AGI play\" sequence on cross — the punchline tweet at `042926TT.txt:2754`.\n- **Box upload:** 2026-04-29 15:37:43 PT — Day 3 late-afternoon Defense batch (uploaded same minute as [[DX 819]] and [[DX 844]]).\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~367 KB, 1 page.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `1444.pdf`.\n\n## Transcribed text\n\n> **Elon Musk** ✓ ✕\n> @elonmusk\n>\n> **Tesla will be one of the companies to make AGI and probably the first to make it in humanoid/atom-shaping form**\n>\n> 1:15 AM · Mar 4, 2026 · 57.3M Views\n>\n> 10K replies | 11K reposts | 132K likes | 6.3K bookmarks\n\n## Commentary\n\nDX 1444 is the **closing line of defense's \"Tesla / xAI as Musk's own AGI play\" theme** — the public declaration eight weeks before trial that **Tesla** (a for-profit Musk controls, with no profit cap, no charitable mission, and no public-benefit governance) intends to be **the first** to AGI in humanoid/atom-shaping form. Defense uses it together with [[DX 844]] (Nov 2018 \"decided to attempt that through Tesla instead\"), DX 749 (Karpathy's \"OpenAI to attach to Tesla as its cash cow,\" with Musk's endorsement), DX 761 (Feb 2018 Musk to Zilis, \"actively try to move three or four people from OpenAI to Tesla\"), DX 853 (Dec 2018 \"zero percent: not 1 percent\"), and Musk's Day 4 cross concession that xAI \"doesn't have any profit caps for its investors\" to argue that Musk's \"stole a charity\" framing is internally inconsistent with his own contemporaneous conduct and his own current commercial AGI ambitions. Plaintiffs' answer on the same exhibit: the tweet does not deny the existence of the OpenAI charitable trust, and Musk having an AGI plan of his own is fully consistent with him being a defrauded donor to a charity that turned for-profit without him. See [[Key Themes]] §\"Tesla / xAI as Musk's own AGI play\" for the full sequence.\n\n---\n*See also:* [[Day 3|Day 3 digest]] · [[DX 844]] · [[DX 819]] · [[Elon Musk]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-502", "exhibit": "DX 502", "party": "Defense", "type": "Email thread", "admitted_trial_day": "Day 4 (April 30, 2026)", "uploaded_box_pt": "2026-04-30T15:13:20", "uploader": "Morrison Foerster", "pages": 3, "size_bytes": 251138, "source_pdf": "DX-502.pdf", "pdf_url": "https://media.mts-in.com/DX-502.pdf", "body_markdown": "# DX 502 — June 24, 2015 Altman → Musk \"AI lab\" five-point email + Musk's \"Agree on all\" reply\n\n> The earliest known full-prose pitch for what would become OpenAI: Sam Altman's June 24, 2015 numbered five-point proposal — mission, 7–10 founding researchers, a five-person foundation board (Musk, Gates, Omidyar, Moskovitz, Altman), part-time-partner role for Musk, and a planned safety regulation letter — and Musk's two-word reply, \"**Agree on all**.\"\n\n## Document type\n**Email thread, plain text, two messages.** Musk's June 24, 2015 11:05:44 PM reply (\"Agree on all\") on top of Altman's same-day 10:24:54 AM original. Subject \"Re: AI lab.\" Bates 2024MUSK-0001230–0001232. Document ID DX-0502.\n\n## Logistics\n- **Trial admission:** Day 4 (April 30, 2026) — used during Musk cross / Birchall direct sequence. No prior wiki reference; admission day inferred from Box upload date.\n- **Box upload:** 2026-04-30 15:13:20 PT — Day 4 mid-afternoon defense batch (clustered with [[DX 545]], [[DX 1156]], [[DX 1157]]).\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~245 KB, 3 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `502 .pdf` (note trailing space).\n\n## Transcribed text\n\n> **From:** Elon Musk\n> **To:** Sam Altman\n> **Subject:** Re: AI lab\n> **Date:** Wednesday, June 24, 2015 11:05:44 PM\n>\n> Agree on all\n>\n> ---\n>\n> **From:** Sam Altman\n> **To:** Elon Musk\n> **Subject:** AI lab\n> **Date:** Wednesday, June 24, 2015 10:24:54 AM\n>\n> > 1) The mission would be to create the first general AI and use it for individual empowerment—ie, the distributed version of the future that seems the safest. More generally, safety should be a first-class requirement.\n> >\n> > 2) I think we'd ideally start with a group of 7-10 people, and plan to expand from there. We have a nice extra building in Mountain View they can have.\n> >\n> > 3) I think for a governance structure, we should start with 5 people and I'd propose you, Bill Gates, Pierre Omidyar, Dustin Moskovitz, and me. The technology would be owned by the foundation and used \"for the good of the world\", and in cases where it's not obvious how that should be applied the 5 of us would decide. The researchers would have significant financial upside but it would be uncorrelated to what they build, which should eliminate some of the conflict (we'll pay them a competitive salary and give them YC equity for the upside). We'd have an ongoing conversation about what work should be open-sourced and what shouldn't. At some point we'd get someone to run the team, but he/she probably shouldn't be on the governance board.\n> >\n> > 4) Will you be involved somehow in addition to just governance? I think that would be really helpful for getting work pointed in the right direction getting the best people to be part of it. Ideally you'd come by and talk to them about progress once a month or whatever. We generically call people involved in some limited way in YC \"part-time partners\" (we do that with Peter Thiel for example, though at this point he's very involved) but we could call it whatever you want. Even if you can't really spend time on it but can be publicly supportive, that would still probably be really helpful for recruiting.\n> >\n> > 5) I think the right plan with the regulation letter is to wait for this to get going and then I can just release it with a message like \"now that we are doing this, I've been thinking a lot about what sort of constraints the world needs for safefy.\" I'm happy to leave you off as a signatory. I also suspect that after it's out more people will be willing to get behind it.\n> >\n> > Sam\n\n## Commentary\n\nDX 502 is the defense's \"Genesis email.\" It is the cleanest single document showing Musk in active *agreement* with the founding architecture: a five-member foundation board including Altman, **technology owned by the foundation and used \"for the good of the world,\"** researcher comp via \"**YC equity for the upside**\" (i.e., for-profit upside vehicle from day one), and a Musk role explicitly framed as \"**part-time partner**\" / \"**publicly supportive**\" rather than CEO. Defense cross uses (i) the **YC-equity researcher comp** to neutralize plaintiffs' \"stole-a-charity\" framing — Musk signed off on a hybrid foundation-+-equity model from the start; (ii) the **part-time partner** language to undercut Musk's claim that he was the dispositive operational founder; and (iii) Musk's blanket \"**Agree on all**\" to argue contemporaneous consent to the eventual structural mix. Plaintiffs' counter-read: the technology was to be **\"owned by the foundation\"** and used \"for the good of the world,\" and the research equity was uncorrelated to what they built — both promises plaintiffs say the [[PX 236|2018 LP term sheet]] and the post-2019 commercial conversion betrayed.\n\n---\n*See also:* [[Day 4|Day 4 digest]] · [[Sam Altman]] · [[PX 7]] · [[PX 25]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-507", "exhibit": "DX 507", "party": "Defendants", "type": "Email thread", "admitted_trial_day": "Day 3 (April 29, 2026)", "uploaded_box_pt": "2026-04-29T15:37:43", "uploader": "Morrison Foerster", "pages": 1, "size_bytes": 363333, "source_pdf": "DX-507.pdf", "pdf_url": "https://media.mts-in.com/DX-507.pdf", "body_markdown": "# DX 507 — Nov 20, 2015 Musk-to-Altman email: \"Probably better to have a standard C corp with a parallel nonprofit\"\n\n> Musk's two-paragraph reply on top of Sam Altman's \"Plan is to have you, me, and Ilya on the Board of Directors for YC AI\" email — the contemporaneous receipt of Musk himself proposing a **standard C corp with a parallel nonprofit** as the structure for what became OpenAI.\n\n## Document type\n**Email thread, plain text, two messages.** Musk's Nov 20, 2015 8:29 PM UTC reply on top of Altman's Nov 20, 2015 11:48 AM email. Subject: \"Re: AI docs.\" From: Elon Musk <erm@spacex.com> to Sam Altman <sam@ycombinator.com>. Bates 2024MUSK-0009963.\n\n## Logistics\n- **Trial admission:** Day 3 (April 29, 2026) — Musk direct concludes; cross begins. Per the chunk wiki note: \"DX 507 — 2015 C-corp with parallel nonprofit email.\" Centerpiece of defense's \"Musk himself wanted a for-profit from the start\" line — see [[Key Themes]] § \"'Unequivocal control' (the 2017 negotiation breakdown)\" and the broader \"tale of two Elons\" framing.\n- **Box upload:** 2026-04-29 15:37:43 PT — Day 3 mid-afternoon batch.\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~355 KB, 1 page.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `507 .pdf`.\n\n## Transcribed text\n\n> **From:** \"Elon Musk\" <erm@spacex.com>\n> **To:** \"Sam Altman\" <sam@ycombinator.com>\n> **Subject:** Re: AI docs\n> **Date:** Fri, 20 Nov 2015 20:29:18 -0000\n> **Importance:** Normal\n>\n> I think this should be independent from (but supported by) YC, not what sounds like a subsidiary.\n>\n> Also, the structure doesn't seem optimal. In particular, the YC stock along with a salary from the nonprofit muddies the alignment of incentives. **Probably better to have a standard C corp with a parallel nonprofit.**\n\n> **From:** Sam Altman <sam@ycombinator.com>\n> **To:** Elon Musk <erm@spacex.com>\n> **Subject:** AI docs\n> **Date:** Fri, Nov 20, 2015 at 11:48 AM\n>\n> Elon--\n>\n> Plan is to have you, me, and Ilya on the Board of Directors for YC AI, which will be a Delaware non-profit. We will also state that we plan to elect two other outsiders by majority vote of the Board.\n>\n> We will write into the bylaws that any technology that potentially compromises the safety of humanity has to get consent of the Board to be released, and we will reference this in the researchers' employment contracts.\n>\n> At a high level, does that work for you?\n>\n> I'm cc'ing our GC Jon Levy here--is there someone in your office he can work with on the details?\n>\n> Sam\n\n## Commentary\n\nDX 507 is the cleanest single document in the defense's \"tale of two Elons\" set: in November 2015, before OpenAI was even publicly announced, **Musk himself proposed the for-profit/nonprofit dual structure** that he would, eight years later, sue OpenAI for adopting. **Defense's read:** the email forecloses any claim that the for-profit conversion was a betrayal of Musk's original vision — Altman in fact proposed a *single nonprofit* with a humanity-safety bylaw veto, and **Musk pushed back to insert a \"standard C corp with a parallel nonprofit.\"** The same month, OpenAI was launched as a single 501(c)(3); the for-profit subsidiary did not arrive until 2019. **Plaintiffs' read:** the November 2015 idea was abandoned at Musk's pre-launch blessing of the nonprofit-only structure; Musk's $38M of subsequent contributions were made to that nonprofit, not to a C corp. The defense pairs this exhibit with [[PX 157]] (the Sept 2017 \"unequivocal control\" negotiation) to argue Musk's objection has always been about *who controls* the for-profit, not *whether* one exists. See [[Day 3|Day 3 digest]] for the cross use, and [[Key Themes]] § \"'Unequivocal control' (the 2017 negotiation breakdown).\"\n\n---\n*See also:* [[Day 3|Day 3 digest]] · [[PX 157]] · [[Sam Altman]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-508", "exhibit": "DX 508", "party": "Defense (OpenAI/MS)", "type": "Text messages", "admitted_trial_day": "Day 6 (May 4, 2026)", "uploaded_box_pt": "2026-05-04T18:15:18", "uploader": "Morrison Foerster", "pages": 3, "size_bytes": 458738, "source_pdf": "DX-508.pdf", "pdf_url": "https://media.mts-in.com/DX-508.pdf", "body_markdown": "# DX 508 — Nov 21, 2015 Brockman / Sutskever pre-OpenAI text thread (\"since he'll be bankrolling it it'll be hard to stop it\")\n\n> A 22-message chat between Greg Brockman and Ilya Sutskever the day before they sat down to dinner with Musk in Woodside — Sutskever worrying out loud that \"since he'll be bankrolling it it'll be hard to stop it,\" Brockman reassuring him that \"I talked about this explicitly with Sam (who I trust on this), I think it'll be fine.\"\n\n## Document type\n**Text messages.** Cellebrite-style \"Short Message Report\" produced from Brockman's phone: a single conversation (\"88J01T\"), 22 messages, 5 participants (Brockman, Sutskever, three redacted), date range 11/21/2015. Bates OPENAI_MUSK00021615–21617. Production-stamped CONFIDENTIAL.\n\n## Logistics\n- **Trial admission:** Day 6 (May 4, 2026) — Russell direct + cross; Brockman cross by Mr. Kry. Used by the defense on Brockman's cross to date the founders' wariness about Musk's \"bankrolling\" leverage to *before* OpenAI was formally announced.\n- **Box upload:** 2026-05-04 18:15:18 PT — same-day upload, in the post-trial-day batch.\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~448 KB, 3 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `0508.pdf`.\n\n## Transcribed text\n\n> **Short Message Report**\n> Conversations: 1 · Participants: 5 · Total Messages: 22 · Date Range: 11/21/2015\n>\n> **88J01T — 11/21/2015 — Chat** · 22 messages on 11/21/2015 · Greg Brockman (owner) · Ilya Sutskever · Me (owner — Greg Brockman) · Greg Brockman · Unknown <owner>\n>\n> *Messages in chronological order (times are shown in GMT +00:00)*\n>\n> **Me (Greg Brockman) — 11/21/2015, 5:10 PM**\n> Ping when you're around?\n>\n> **Ilya Sutskever — 11/21/2015, 5:54 PM**\n> Hi\n>\n> **Ilya Sutskever — 11/21/2015, 5:54 PM**\n> I'm on my way home, could talk in 20 minutes\n>\n> **Me (Greg Brockman) — 11/21/2015, 5:54 PM**\n> sg\n>\n> **Ilya Sutskever — 11/21/2015, 6:17 PM**\n> I'm ready\n>\n> **Me (Greg Brockman) — 11/21/2015, 6:17 PM**\n> cool. on a call but will call after (5-10 mins)\n>\n> **Ilya Sutskever — 11/21/2015, 6:17 PM**\n> No problems\n>\n> **Ilya Sutskever — 11/21/2015, 7:08 PM**\n> BTW I am happy to go to SF for tomorrow's dinner, especially if you can think of a nice restaurant\n>\n> **Me (Greg Brockman) — 11/21/2015, 7:23 PM**\n> thanks! actually, already have a great restaurant in Woodside in mind — village pub (http://thevillagepub.net/). they have my favorite desert in the world. 7:45p work?\n>\n> **Ilya Sutskever — 11/21/2015, 7:25 PM**\n> Yes, excellent! See you tomorrow!\n>\n> **Me (Greg Brockman) — 11/21/2015, 7:25 PM**\n> great, see you then!\n>\n> **Ilya Sutskever — 11/21/2015, 8:27 PM**\n> Elon might spend half a day a week with us\n>\n> **Ilya Sutskever — 11/21/2015, 9:09 PM**\n> I imagined how it will be and I worry that our work environment can become very stressful\n>\n> **Ilya Sutskever — 11/21/2015, 9:11 PM**\n> And since he'll be bankrolling it it'll be hard to stop it\n>\n> **Me (Greg Brockman) — 11/21/2015, 9:12 PM**\n> I talked about this explicitly with Sam (who I trust on this), I think it'll be fine. Talking to Elon tomorrow, will let you know how it goes\n>\n> **Ilya Sutskever — 11/21/2015, 9:12 PM**\n> Ok LMK how it goes\n>\n> **Me (Greg Brockman) — 11/21/2015, 9:13 PM**\n> (In general I think these things are mostly a mix of setting expectations and having good working relationships. So pretty addressable if you know what you're worried about, and sometimes even if you don't know.)\n>\n> **Ilya Sutskever — 11/21/2015, 9:14 PM**\n> Sounds reasonable\n>\n> **Ilya Sutskever — 11/21/2015, 10:31 PM**\n> In you call with Elon tomorrow, be sure to remind him that research takes longer than one might expect, and that deepmind is a relaxed environment\n>\n> **Me (Greg Brockman) — 11/21/2015, 10:32 PM**\n> For sure.\n>\n> **Me (Greg Brockman) — 11/21/2015, 10:33 PM**\n> Don't stress about it. It'll be fine!\n>\n> **Ilya Sutskever — 11/21/2015, 10:33 PM**\n> I hope so!\n\n## Commentary\n\nDX 508 is the earliest contemporaneous record in the trial of the founders worrying about Musk's *control* — and it predates OpenAI's December 2015 launch by about three weeks. Defense uses Sutskever's line \"**since he'll be bankrolling it it'll be hard to stop it**\" to argue that the \"absolute control\" worry plaintiffs say emerged only in the September 2017 [[PX 157]] negotiation was actually baked into the founders' dinner-prep texts before the nonprofit existed. Brockman's reply — that he had \"talked about this explicitly with Sam (who I trust on this)\" — is also defense-favorable: it puts Altman in the role of *counterbalance* to Musk from day one, which dovetails with [[PX 157]]'s \"Sam acted as a genuine counterbalance to you, which has been extremely fruitful.\" The exhibit pairs naturally with the [[Brockman Journal]] entries from 2017 that re-tread the same anxiety.\n\n---\n*See also:* [[Day 6|Day 6 digest]] · [[Greg Brockman]] · [[Ilya Sutskever]] · [[PX 157]] · [[Brockman Journal]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-509", "exhibit": "DX 509", "party": "Defense", "type": "Email thread", "admitted_trial_day": "Day 6 (May 4, 2026)", "uploaded_box_pt": "2026-05-04T18:15:18", "uploader": "Morrison Foerster", "pages": 27, "size_bytes": 2254085, "source_pdf": "DX-509.pdf", "pdf_url": "https://media.mts-in.com/DX-509.pdf", "body_markdown": "# DX 509 — Nov 22–24, 2015 Brockman ↔ Musk ↔ Altman thread on the OpenAI launch: $1B funding commitment, \"Cogito\"/\"Freemind\" naming, $175k salaries, headcount caps\n\n> The four-day pre-launch email exchange (Nov 22–24, 2015) in which Brockman, Musk and Altman work out the **$1B funding commitment** language (\"This is real. I will cover whatever anyone else doesn't provide\"), the candidate names for the new entity (Axon, AI Summer, Difference Engine, Cogito, Freemind), the **$175k base salary + $1.5M five-year cash bonus** offer template, and the founding-period staff-growth caps (\"Until we see a road to AGI, we should almost certainly not scale beyond 120 people\").\n\n## Document type\n**Email thread, plain text, multiple messages** between Greg Brockman, Elon Musk, and Sam Altman (cc Sam Altman / sam@ycombinator.com) over Nov 22–24, 2015. Subject: \"follow up from call\" / \"Re: follow up from call.\" Bates 2024MUSK-0000384–0000410.\n\n## Logistics\n- **Trial admission:** Day 6 (May 4, 2026) — Brockman cross by Mr. Kry. Defense exhibit anchoring the founding-period record of how the $1B figure, the structure, and the recruiting comp template were set, weeks before the public Dec 11, 2015 launch.\n- **Box upload:** 2026-05-04 18:15:18 PT — Day 6 late-afternoon Defense batch.\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~2.2 MB, 27 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `0509.pdf`.\n\n## Transcribed text\n\n> **From:** Elon Musk\n> **To:** Greg Brockman\n> **Cc:** Sam Altman\n> **Subject:** Re: follow up from call\n> **Date:** Tuesday, November 24, 2015 2:09:53 AM\n>\n> I think we need a very simple comp structure, which you pretty much have in the offer letters, apart from how one should value the stock of YC companies. At the risk of sounding pessimistic, I think we are going to see a lot of startup death in the next few years and the value of equity in general will drop massively. We are already in the early part of the down cycle.\n>\n> In order to know what comp makes sense, it would be helpful to understand what comp each candidate is being offered by other companies. We don't have to match Google, but we also can't be an order of magnitude off. Ilya was very clear that the current offer wouldn't cut it. I don't know how the others feel, as they didn't say anything explicitly and I haven't talked to everyone.\n>\n> Despite what SpaceX is doing, since this is located in the valley, I'd go with straight four vesting evenly distributed. That's what we do at Tesla to match market conditions. However, make the initial cash vest grant much lower and commit to do anywhere from 50% to 200%+ of the initial grant every year, depending on whether they have an ok or an amazing year. These would naturally stack over time.\n>\n> If we don't have differential comp, we risk losing at least some great people as soon as they do something notable and get a giant offer from the usual suspects. They will blame on their spouse :~\n>\n> We absolutely shouldn't differentiate between engineers and \"researchers\". The latter is just an engineer with ego issues. SpaceX is at the absolute forefront of the advancement of space — far beyond what any so-called research professor is doing — but no one is called anything but an engineer.\n>\n> I have a crazy packed day tomorrow, but am free to talk on Wed.\n\n> **From:** Elon Musk\n> **To:** Greg Brockman\n> **Cc:** Sam Altman\n> **Subject:** Re: follow up from call\n> **Date:** Monday, November 23, 2015 4:33:24 PM\n>\n> Not bad. Sounds kinda cute. Most people won't get the latin, but the ones we want to join will.\n>\n> I'd support that.\n\n> **From:** Greg Brockman\n> **To:** Elon Musk\n> **Cc:** Sam Altman\n> **Subject:** Re: follow up from call\n> **Date:** Monday, November 23, 2015 4:33:24 PM\n>\n> > Oof, good catch on Exxon.\n> >\n> > Any reaction to \"Cogito\"? Sam and I were just discussing and like it a lot — has a nice individualistic feel to it.\n> > - gdb\n\n> **From:** Elon Musk\n> **To:** Greg Brockman\n> **Cc:** Sam Altman\n> **Subject:** RE: follow up from call\n> **Date:** Monday, November 23, 2015 4:10:00 PM\n> > >\n> > > **Axon is the best of the three, but I don't love it. A bit too biological and saying it out loud sounds a lot like \"Exxon\", one of the worst companies on Earth.** I'm going back and forth on \"Freemind\". Kinda like \"Freeman\" too, as it reminds me of the scientist protagonist in Half-Life, who was an awesome character, and it sounds like what we are essentially trying to achieve, which is **maximum freedom of action for humanity**.\n> > >\n> > > **The billion would be a funding commitment over ten years (more if necessary), but we don't need to mention the ten year part externally. Totally agree that this should not be a license to overspend. The operating standard should be one of frugality. Even at a billion dollars, we will be massively outmanned and outgunned by Google and Facebook, but we will have right on our side and that counts for a lot.**\n> > >\n> > > The SpaceX cash vesting bonus is anywhere from 0% (for a needs improvement rating) to 300% (mind-blowing ass-kicker) over 5 years. This was a straight vesting schedule, but we are changing it to 10, 15, 20, 25, 30. Default is cash, though you can choose equiv amount in stock or 2X the number in options. … We don't want someone thinking or their spouse telling them that they could have done vastly better somewhere else.\n\n> > > > **From:** Greg Brockman — **Sent:** Monday, November 23, 2015 3:35 PM:\n> > > >\n> > > > > Blog sounds good, assuming adjustments for neutrality vs being YC-centric.\n> > > >\n> > > > Yep. On the name front, the best three Sam and I brainstormed last night:\n> > > > - Axon\n> > > > - AI Summer\n> > > > - Difference Engine\n> > > >\n> > > > Curious if any of those appeal (no worries if you think they are all bad :)).\n> > > >\n> > > > > I'd favor positioning the blog to appeal a bit more to the general public — there is a lot of value to having the public root for us to succeed — and then having a longer, more detailed and inside-baseball version for recruiting, with a link to it at the end of the general public version.\n> > > >\n> > > > Yep, fair. I'll try drafting something along those lines today to get a feel for it.\n> > > >\n> > > > > **We need to go with a much bigger number than $100M to avoid sounding hopeless relative to what Google or Facebook are spending. I think we should say that we are starting with a $1B funding commitment. This is real. I will cover whatever anyone else doesn't provide.**\n> > > >\n> > > > For what it's worth, I was talking to Ilya last night, and he also expressed concern that Google is going to be deploying ever-increasing amounts of cash to get people (they have a very clear ROI on deep learning research). … I do think we could make a play with $100M, but with $1B we remove a large risk factor.\n> > > >\n> > > > So, I'm on board. We just need to be very mindful, have the same culture of frugality we would have otherwise, and acknowledge we still need to retain people by building the best environment rather than trying to match Google's offers.\n\n> **From:** Greg Brockman\n> **To:** Elon Musk\n> **Cc:** Sam Altman\n> **Subject:** Re: follow up from call\n> **Date:** Tuesday, November 24, 2015 1:14:22 AM\n>\n> [Brockman's compensation-framework memo to Musk and Altman sets out the structured questions that drive the rest of the thread:]\n>\n> \"What's the right total 5-year compensation package?\" — proposes **$175k salary + $1.5M cash bonus** vesting on the SpaceX schedule (Year 1: $150k bonus / $325k total; Year 2: $225k / $400k; Year 3: $300k / $475k; Year 4: $375k / $550k; Year 5: $450k / $625k).\n>\n> \"Is there enough wiggle room for performance?\" / \"Can people later choose a different option (cash vs YC stock vs SpaceX stock)?\" / \"How do we reconcile a 5-year vesting cash bonus with the current offer to receive YC stock indefinitely?\" / \"How do we expect future people's compensation to compare to that of the founding team?\" — \"in a (successful) startup, the early people get compensated way more highly than others. … I think we should declare the current offer the 'senior' compensation level… I think we should make an exception for truly exceptional and unique people like Ilya — call that the 'super senior', custom compensation level — which would probably be a small whitelist of individuals like Geoff Hinton or Sergey Levine.\"\n>\n> \"How will we handle future giant-offer-from-Google situations?\" / \"Should we be including engineers in the 'senior' compensation level?\" — \"Market rate for engineers is obviously lower than that for deep learning experts. I suspect we should have few engineers in the 'senior' level. But that being said: 1) culturally, I think we want a minimal in formal distinctions between engineers and researchers, and I expect the lines to blur a lot, 2) the founding group should all be equals, 3) we should make sure to culturally value good engineering to the same degree as good research.\"\n>\n> \"What other levels do we expect to exist?\" — research intern tier; \"What do we expect headcount growth to look like? My thoughts: Right now we have a total of 10 full-time members. I'm most worried about growing the team too quickly and losing the specialness. (I've heard this happened to DeepMind.) … After 12 months… 15-20 permanent staff, 5-10 interns. After 24 months… 25-40 permanent staff, 10-15 interns. **Until we see a road to AGI, we should almost certainly not scale beyond 120 people, nor more than 2x year over year.**\"\n\n> **From:** Greg Brockman\n> **To:** Elon Musk\n> **Cc:** Sam Altman\n> **Subject:** follow up from call\n> **Date:** Sunday, November 22, 2015 6:11:13 PM\n> **Attachments:** YCR offer letter (template).pdf\n> *[earliest message in thread]*\n>\n> \"Hey Elon, Nice chatting earlier. As I mentioned on the phone, here's the latest early draft of the blog post: https://quip.com/6YnqA26RJgKr. (Sam, Ilya, and I are thinking about new names; would love any input from you.) … I think it's mostly important that our messaging appeals to the research community (or at least the subset we want to hire). **I hope for us to enter the field as a neutral group, looking to collaborate widely and shift the dialog towards being about humanity winning rather than any particular group or company.** (I think that's the best way to bootstrap ourselves into being a leading research institution.) I've attached the offer letter template we've been using, with a salary of $175k. Here's the email template I've been sending people: 'Attached is your official YCR offer letter! … There will be two more documents coming: A separate letter offering you 0.25% of each YC batch you are present for (as compensation for being an Advisor to YC). The At-Will Employment, Confidential Information, Invention Assignment and Arbitration Agreement … We plan to offer the following benefits: Health, dental, and vision insurance; Unlimited vacation days with a recommendation of four weeks per year; Paid parental leave; Paid conference attendance when you are presenting YC AI work or asked to attend by YC AI. We're also happy to provide visa support. When you're ready to talk about visa-related questions, I'm happy to put you in touch with Kirsty from YC.'\"\n\n## Commentary\n\nDX 509 is defense's clean record of **how the $1B figure was actually constructed** — Musk's \"**The billion would be a funding commitment over ten years (more if necessary), but we don't need to mention the ten year part externally**\" line is the documentary backbone for defense's \"$1B / $38M gap\" point on cross (see [[Key Themes]] §\"$1B / $38M gap\"). The \"I will cover whatever anyone else doesn't provide\" line cuts the other way for plaintiffs, who use it to argue Musk personally guaranteed the funding. The thread also previews the **OPEN-AI naming choice** (the rejected candidates \"Axon,\" \"AI Summer,\" \"Difference Engine,\" \"Cogito,\" \"Freemind\") and pegs Musk's stated motivation to \"**maximum freedom of action for humanity**.\" The compensation memo is the original of the recruiting practice that Brockman would later defend at trial, and the **\"Until we see a road to AGI, we should almost certainly not scale beyond 120 people\"** ceiling is a useful defense data point against any narrative that the founders were planning a for-profit empire from the start.\n\n---\n*See also:* [[Day 6|Day 6 digest]] · [[PX 14]] · [[PX 5]] · [[Greg Brockman]] · [[Sam Altman]] · [[Elon Musk]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-516", "exhibit": "DX 516", "party": "Defense", "type": "Email thread", "admitted_trial_day": "Day 4 (April 30, 2026)", "uploaded_box_pt": "2026-04-30T15:13:20", "uploader": "Morrison Foerster", "pages": 2, "size_bytes": 613795, "source_pdf": "DX-516.pdf", "pdf_url": "https://media.mts-in.com/DX-516.pdf", "body_markdown": "# DX 516 — Jan 2, 2016 Sutskever → Musk, \"less open\" (Demis Hassabis Falcon-9 forward); Musk reply \"Yup\"\n\n> A two-page email thread from January 2, 2016 in which Demis Hassabis (DeepMind/Google) congratulates Musk on the Falcon 9 landing and questions the open-sourcing strategy as a safety panacea — Sutskever forwards Musk a written defense conceding \"**As we get closer to building AI, it will make sense to start being less open**\" — and Musk replies \"**Yup.**\"\n\n## Document type\n**Email thread, plain text, three-message chain.** Musk's \"Yup\" reply on top of Sutskever's analysis reply on top of a Demis Hassabis \"congrats on the falcon 9\" forward. To: Ilya Sutskever; cc: sam@ycombinator.com, Greg Brockman. Subject \"Re: congrats on the falcon 9.\" Bates 2024MUSK-0005755–5756.\n\n## Logistics\n- **Trial admission:** Day 4 (April 30, 2026 — Musk cross + redirect; Birchall direct + cross). Per [[Key Themes|Day 4 quotes table]]: \"DX 516 — Sutskever email 'as we get closer to building AI, it will make sense to start being less open'; Musk: 'Yup'\" `(043026TT.txt:1396–1410)`.\n- **Box upload:** 2026-04-30 15:13:20 PT — Day 4 mid-afternoon defense batch (clustered with DX 539, DX 1156, DX 1285, DX 849, DX 1083).\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~599 KB, 2 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `516 .pdf` (note trailing space in filename).\n\n## Transcribed text\n\n> **From:** Elon Musk <erm@spacex.com>\n> **To:** Ilya Sutskever <ilyasu@openai.com>\n> **Cc:** Sam Altman <sam@ycombinator.com>, Greg Brockman <gdb@openai.com>\n> **Subject:** Re: congrats on the falcon 9\n> **Date:** Sat, 02 Jan 2016 17:11:18 -0000\n>\n> Yup\n\n> **From:** Ilya Sutskever <ilyasu@openai.com>\n> **To:** Elon Musk <erm@spacex.com>\n> **Cc:** Sam Altman <sam@ycombinator.com>, Greg Brockman <gdb@openai.com>\n> **Subject:** Re: congrats on the falcon 9\n> **Date:** Sat, Jan 2, 2016 at 11:06 AM\n>\n> The article is concerned with a hard takeoff scenario: if a hard takeoff occurs, and a safe AI is harder to build than an unsafe one, then by opensorucing everything, we make it easy for someone unscrupulous with access to overwhelming amount of hardware to build an unsafe AI, which will experience a hard takeoff.\n>\n> **As we get closer to building AI, it will make sense to start being less open. The Open in openAI means that everyone should benefit from the fruits of AI after its built, but it's totally OK to not share the science** (even though sharing everything is definitely the right strategy in the short and possibly medium term for recruitment purposes).\n\n> **From:** Elon Musk <erm@spacex.com>\n> **To:** Ilya Sutskever <ilyasu@openai.com>\n> **Cc:** Sam Altman <sam@ycombinator.com>, Greg Brockman <gdb@openai.com>\n> **Subject:** Fwd: congrats on the falcon 9\n> **Date:** Sat, Jan 2, 2016 at 8:18 AM\n>\n> Begin forwarded message:\n\n> **From:** Demis Hassabis <demishassabis@google.com>\n> **To:** Elon Musk <erm@spacex.com>\n> **Subject:** congrats on the falcon 9\n> **Date:** January 2, 2016 at 10:12:32 AM CST\n>\n> Hi Elon\n> >\n> > Happy new year to you, Talulah and the boys!\n> >\n> > Congratulations on landing the Falcon 9, what an amazing achievement. Time to build out the fleet now!\n> >\n> > I've seen you (and Sam and other OpenAI people) doing a lot of interviews recently extolling the virtues of open sourcing AI, but I presume you realise that this is not some sort of panacea that will somehow magically solve the safety problem? There are many good arguments as to why the approach you are taking is actually very dangerous and in fact may increase the risk to the world. Some of the more obvious points are well articulated in this blog post, that I'm sure you've seen, but there are also other important considerations:\n> >\n> > http://slatestarcodex.com/2015/12/17/should-ai-be-open/\n> >\n> > I'd be interested to hear your counter-arguments to these points.\n> >\n> > Best\n> > Demis\n\n## Commentary\n\nDX 516 is the defense's cleanest **\"OPEN\" wasn't always open-source** receipt — three weeks after OpenAI's launch, Sutskever is on the record explaining to Musk (and Musk endorsing with \"**Yup.**\") that \"**The Open in openAI means that everyone should benefit from the fruits of AI after its built, but it's totally OK to not share the science**\" and that \"**As we get closer to building AI, it will make sense to start being less open.**\" Defense uses this against the [[Key Themes|\"OPEN\" in OpenAI]] direct examination (\"the 'OPEN' in OpenAI represents open source\") to argue Musk was internally aware of, and *agreed with*, the eventual closed-science trajectory back in January 2016 — i.e., the same trajectory that crested in the [[PX 251|2020 Microsoft GPT-3 exclusive license]] he later tweeted was \"the opposite of open.\" It is also one of the only Musk → Sutskever exchanges in evidence that pre-dates the August 2017 control-fight period. The Demis Hassabis cameo is incidental but useful texture: DeepMind's CEO was prodding the OpenAI principals on safety months before the [[Key Themes|\"speciesist\" anecdote]] became Musk's public origin story. See [[Day 4|Day 4 digest]] at 043026TT.txt:1396–1410 for the cross use.\n\n---\n*See also:* [[Day 4|Day 4 digest]] · [[Ilya Sutskever]] · [[Elon Musk]] · [[Demis Hassabis]] · [[PX 251]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-521", "exhibit": "DX 521", "party": "Defendants", "type": "Email thread", "admitted_trial_day": "Day 6 (May 4, 2026)", "uploaded_box_pt": "2026-05-04T18:15:17", "uploader": "Morrison Foerster", "pages": 3, "size_bytes": 304616, "source_pdf": "DX-521.pdf", "pdf_url": "https://media.mts-in.com/DX-521.pdf", "body_markdown": "# DX 521 — Feb 18 – Mar 1, 2016 OpenAI–Musk weekly-visit scheduling email thread (incl. Sutskever's \"Regular infrequent meetings with Elon\" + Musk \"Demis dinner / Super Bowl / Puppy Bowl\")\n\n> Reverse-chron defense email thread showing Sutskever asking for biweekly meetings with Musk, Musk's Feb 18, 2016 \"**My dinner with Demis was extremely alarming. I feel like they are playing the Super Bowl and we are playing the Puppy Bowl**\" reaction email, and the OpenAI team scheduling weekly Tuesday office visits — defense's documentary support for the early-2016 \"tightly engaged Musk\" period.\n\n## Document type\n**Email thread, multi-message reverse-chronological chain.** Top message dated Mar 1, 2016 8:28 AM from Sam Teller (steller@spacex.com) to Sam Altman, Emma Gallagher, Greg Brockman, Ilya Sutskever. Bates OPENAI_MUSK00029161 through OPENAI_MUSK00029163. Marked Confidential.\n\n## Logistics\n- **Trial admission:** Day 6 (May 4, 2026) — Russell direct + cross; Brockman cross by Mr. Kry.\n- **Box upload:** 2026-05-04 18:15:17 PT (defense upload window).\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~298 KB, 3 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `0521.pdf`.\n\n## Transcribed text\n\n> **From:** Sam Teller <steller@spacex.com>\n> **To:** Sam Altman <sam@ycombinator.com>, Emma Gallagher <emma@spacex.com>\n> **Cc:** Greg Brockman <gdb@openai.com>, Ilya Sutskever <ilyasu@openai.com>\n> **Subject:** Re: OpenAI - Weekly Visit\n> **Date:** 3/1/2016 8:28:10 AM\n>\n> Hey guys,\n>\n> Due to the SpaceX launch tomorrow we have to cancel Elon's visit. Sorry about the last minute notice. We'll come to the office next week.\n>\n> Thanks,\n> Sam\n\n> **From:** Sam Altman <sam@ycombinator.com>\n> **To:** Emma Gallagher <emma@spacex.com>\n> **Cc:** Greg Brockman <gdb@openai.com>, Ilya Sutskever <ilyasu@openai.com>, Sam Teller <steller@spacex.com>\n> **Subject:** Re: OpenAI - Weekly Visit\n> **Date:** Friday, February 19, 2016 at 8:24 AM\n>\n> 845 works for me next week and 10-1130 is good ongoing.\n>\n> Sam\n\n> **From:** Emma Gallagher <emma@spacex.com>\n> **To:** Sam Altman <sam@ycombinator.com>\n> **Cc:** Greg Brockman <gdb@openai.com>, Ilya Sutskever <ilyasu@openai.com>, Sam Teller <steller@spacex.com>\n> **Subject:** Re: OpenAI - Weekly Visit\n> **Date:** Fri, Feb 19, 2016 at 1:19 AM\n>\n> Great, just let me know on the appointment.\n>\n> Sam A - ok for your schedule as well?\n>\n> Thanks,\n> Emma\n\n> **From:** Greg Brockman <gdb@openai.com>\n> **To:** Emma Gallagher <emma@spacex.com>\n> **Cc:** Ilya Sutskever <ilyasu@openai.com>, Sam Teller <steller@spacex.com>, Sam Altman <sam@ycombinator.com>\n> **Subject:** Re: OpenAI - Weekly Visit\n> **Date:** 2/18/16, 9:34 PM\n>\n> Yep, that should work well from our end. Let me check with one team member who has a recurring appointment that day (I suspect it's moveable).\n>\n> And 8:45am next Tuesday sounds fine to me!\n>\n> - gdb\n\n> **From:** Emma Gallagher <emma@spacex.com>\n> **To:** Greg Brockman <gdb@openai.com>, Ilya Sutskever <ilyasu@openai.com>, Sam Teller <steller@spacex.com>, Sam Altman <sam@ycombinator.com>\n> **Subject:** OpenAI - Weekly Visit\n> **Date:** Thu, Feb 18, 2016 at 6:05 PM\n>\n> Hi All -\n>\n> Elon would like to try and spend about 1.5 hrs at the office each week. We're thinking that 10-11:30am on Tuesdays will work well most weeks. How does that work for the team?\n>\n> For next week we'll need to stick with a call because Elon's schedule is a bit tough. Would a call at 8:45am be too early for everyone?\n>\n> Thanks,\n> Emma\n\n> **From:** Emma Gallagher <emma@spacex.com>\n> **To:** Ilya Sutskever <ilyasu@openai.com>, Sam Altman <sam@ycombinator.com>, Sam Teller <steller@spacex.com>\n> **Cc:** Greg Brockman <gdb@openai.com>\n> **Subject:** Re: TODAY - 5:00 PM Call\n> **Date:** Thursday, February 18, 2016 at 2:50 PM\n>\n> Great, thanks!\n\n> **From:** Ilya Sutskever <ilyasu@openai.com>\n> **To:** Emma Gallagher <emma@spacex.com>, Sam Altman <sam@ycombinator.com>, Sam Teller <steller@spacex.com>\n> **Cc:** Greg Brockman <gdb@openai.com>\n> **Subject:** Re: TODAY - 5:00 PM Call\n> **Date:** Thursday, February 18, 2016 at 2:47 PM\n>\n> Yes, that works.\n\n> **From:** Emma Gallagher <emma@spacex.com>\n> **To:** Ilya Sutskever <ilyasu@openai.com>, Sam Altman <sam@ycombinator.com>, Sam Teller <steller@spacex.com>\n> **Cc:** Greg Brockman <gdb@openai.com>\n> **Subject:** TODAY - 5:00 PM Call\n> **Date:** Thu, Feb 18, 2016 at 2:24 PM\n>\n> Hi All,\n>\n> Does 5:00pm today work well? If so, below is a conference number we can use.\n>\n> If you can let me know the names of who will be joining that would be great to have.\n>\n> Conference Line 1: +1-855-282-6330 US TOLL FREE / +1-415-655-0003 US TOLL\n> Access Code 1: 28848741 # # *no host needed\n> One-Touch Dial: 18552826330p28848741\n\n> **From:** Sam Altman <sam@ycombinator.com>\n> **To:** Sam Teller <steller@spacex.com>\n> **Cc:** Greg Brockman <gdb@openai.com>, Emma Gallagher <emma@spacex.com>, Ilya Sutskever <ilyasu@openai.com>\n> **Subject:** Re: TODAY - Re: Regular infrequent meetings with Elon\n> **Date:** Thursday, February 18, 2016 at 1:57 PM\n>\n> I'm on a plane until 4:40 but feel free to do it without me, or I could join after that.\n\n> **From:** Sam Teller <steller@spacex.com>\n> **To:** Greg Brockman <gdb@openai.com>\n> **Cc:** Emma Gallagher <emma@spacex.com>, Ilya Sutskever <ilyasu@openai.com>\n> **Subject:** Re: TODAY - Re: Regular infrequent meetings with Elon\n> **Date:** Thu, Feb 18, 2016 at 1:46 PM\n>\n> Emma, maybe 4pm?\n\n> **From:** Greg Brockman <gdb@openai.com>\n> **To:** Sam Teller <steller@spacex.com>\n> **Cc:** Sam Altman <sam@ycombinator.com>, Emma Gallagher <emma@spacex.com>, Ilya Sutskever <ilyasu@openai.com>\n> **Subject:** Re: Regular infrequent meetings with Elon\n> **Date:** Thu, Feb 18, 2016 at 12:35 PM\n>\n> +sama\n>\n> We should be able to do today — Ilya and I are interviewing a candidate 2-2:45 but are otherwise free.\n>\n> - gdb\n\n> **From:** Elon Musk <erm@spacex.com>\n> **To:** Sam Teller <steller@spacex.com>, Greg Brockman <gdb@openai.com>, Ilya Sutskever <ilyasu@openai.com>\n> **Subject:** Re: Regular infrequent meetings with Elon\n> **Date:** Thu, Feb 18, 2016 at 11:46 AM\n>\n> **I think I should probably do this every week. My dinner with Demis was extremely alarming. I feel like they are playing the Super Bowl and we are playing the Puppy Bowl. Unless we want to have our ass handed to us, we need to step up our game dramatically.**\n>\n> Let's have a conf call later today with the key people of OpenAI later today, if possible, otherwise tomorrow.\n\n> **From:** Ilya Sutskever <ilyasu@openai.com>\n> **To:** Sam Teller <steller@spacex.com>, Elon Musk <erm@spacex.com>, Greg Brockman <gdb@openai.com>\n> **Subject:** Regular infrequent meetings with Elon\n> **Date:** Thursday, February 18, 2016 11:28 AM\n>\n> Hi Sam,\n>\n> Greg and I think that **it will be useful for OpenAI's long term success if the two of us had a 30 minute meeting with Elon every two weeks**. Elon has built incredible organizations before, and the more of his lessons and experience could be applied to OpenAI, the more effective we will be.\n>\n> Please let me know if it is possible.\n>\n> Ilya\n\n## Commentary\n\nDX 521 is a defense-uploaded counter to the plaintiffs' \"Musk-was-checked-out\" framing for early 2016: the chain begins with Sutskever proactively asking for biweekly meetings — citing Musk's track record building \"incredible organizations\" — and culminates in Musk's \"**My dinner with Demis was extremely alarming**\" / \"**Super Bowl / Puppy Bowl**\" reply that converts the cadence into *weekly* 1.5-hour office visits. Defense uses this to anchor the [[Key Themes|\"unequivocal control\"]] thread for 2016: Musk was, on this evidence, deeply engaged with OpenAI operations, dictating cadence, and using SpaceX-launch slips as the only valid reason to cancel — an active principal, not a passive donor. Plaintiffs would say the same email confirms their narrative — Musk paying for and steering OpenAI personally during phase one, before the [[PX-157|Sept 20, 2017 final straw]]. The \"Demis\" reference is to **Demis Hassabis** (DeepMind/Google) — the dinner that anchors Musk's [[Key Themes|Larry Page \"speciesist\"]] AGI-safety framing across his trial testimony. See [[Day 6|Day 6 digest]] for the live admission.\n\n---\n*See also:* [[Day 6|Day 6 digest]] · [[Greg Brockman]] · [[Ilya Sutskever]] · [[Sam Altman]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-532", "exhibit": "DX 532", "party": "Defense (OpenAI/MS)", "type": "Email thread", "admitted_trial_day": "Day 4 (April 30, 2026)", "uploaded_box_pt": "2026-04-30T15:13:26", "uploader": "Morrison Foerster", "pages": 1, "size_bytes": 368403, "source_pdf": "DX-532.pdf", "pdf_url": "https://media.mts-in.com/DX-532.pdf", "body_markdown": "# DX 532 — May 21, 2016 Musk reply on Brockman's \"google policy conversation\"\n\n> Greg Brockman flags to Musk, Altman, Sutskever and Sam Teller that Google's policy people want to speak with him — they appear to be \"afraid\" OpenAI will attack them in the media, pressure them to release data or models, and \"build a public narrative that it's wrong to have any closed-open source AI.\" Brockman: \"**our mission is to maximally benefit the world; and we don't have a problem with people keeping things proprietary — it's fine to make money off this stuff, and we may even generate revenue ourselves one day.**\" Musk: \"Whoa, that's really interesting. Who called from Google?\"\n\n## Document type\n**Email thread, 1 page, two messages.** (1) Greg Brockman <gdb@openai.com> to Musk, May 20, 2016 7:35 PM, \"google policy conversation,\" cc Sam Teller / Sam Altman / Ilya Sutskever; (2) Elon Musk <elon@openai.com> reply, May 21, 2016 3:04:00 AM, \"Re: google policy conversation,\" same recipients. Bates OPENAI_MUSK00000089. Stamped CONFIDENTIAL. Document ID DX-0532.\n\n## Logistics\n- **Trial admission:** Day 4 (April 30, 2026) — Musk cross + redirect; Birchall direct + cross. Per the wiki TSV: \"**DX 532 — May 2016 Google policy email.**\"\n- **Box upload:** 2026-04-30 15:13:26 PT — Day 4 mid-afternoon batch.\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~360 KB, 1 page.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `532 .pdf`.\n\n## Transcribed text\n\n> **From:** Elon Musk <elon@openai.com>\n> **To:** Greg Brockman <gdb@openai.com>\n> **Cc:** Sam Teller <steller@openai.com>, Sam Altman <sam@ycombinator.com>, Ilya Sutskever <ilyasu@openai.com>\n> **Subject:** Re: google policy conversation\n> **Date:** 5/21/2016 3:04:00 AM\n>\n> Whoa, that's really interesting. Who called from Google?\n>\n> On a separate note, I spoke with Atit (building owner) about the Stripe space. Sounds like he will come back to us with a more reasonable request.\n\n> **From:** Greg Brockman <gdb@openai.com>\n> **To:** Elon Musk <elon@openai.com>\n> **Cc:** Sam Teller <steller@openai.com>, Sam Altman <sam@ycombinator.com>, Ilya Sutskever <ilyasu@openai.com>\n> **Subject:** google policy conversation\n> **Date:** Fri, May 20, 2016 at 7:35 PM\n>\n> Google's policy people want to speak with me. Sounds like they are afraid that:\n> >\n> > - We are going to attack them in the media and pressure them to release data or models, and\n> > - We'll build a public narrative that it's wrong to have any closed-open source AI\n> >\n> > I don't see obvious benefit in us doing these things, though maybe they could be a tool against DeepMind.\n> >\n> > I plan to say that I can't see why we would do that; our mission is to maximally benefit the world; and we don't have a problem with people keeping things proprietary — it's fine to make money off this stuff, and we may even generate revenue ourselves one day.\n> >\n> > (LMK if you have alternate thoughts.)\n> >\n> > - gdb\n\n## Commentary\n\nDX 532 is a small but useful contemporaneous artifact for defense's [[Key Themes#The \"OPEN\" in OpenAI|\"OPEN\" in OpenAI]] counter-position: in May 2016 — six months after launch — Brockman is telling Google, with Musk, Altman and Sutskever in the loop, that OpenAI's \"**mission is to maximally benefit the world**\" and that **closed/proprietary AI is \"fine\"** (\"we don't have a problem with people keeping things proprietary — it's fine to make money off this stuff, and we may even generate revenue ourselves one day\"). That last clause — \"**we may even generate revenue ourselves one day**\" — is exactly the kind of contemporaneous founder-statement defense uses to neutralize plaintiffs' theory that \"open\" was load-bearing or that any commercial path was off the table at the founding (compare Musk on direct: \"**The 'OPEN' in OpenAI represents open source**,\" `042826TT.txt:5163`; and the judge's open-source admonition to OpenAI counsel after openings, `042826TT.txt:6296`). Musk's reply (\"**Whoa, that's really interesting**\") makes him a copy-recipient who registered no objection. Used on Day 4 during the second half of Musk cross, this exhibit complements [[DX-758]]'s Tesla/AGI scenario set and the [[Key Themes#The 2018 \"purple box\" term sheet|purple-box]] / \"infamous purple box\" sequence as evidence that the founders' public messaging in 2016–2018 was already contemplating revenue and proprietary structure.\n\n---\n*See also:* [[Day 4|Day 4 digest]] · [[Greg Brockman]] · [[Elon Musk]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-539", "exhibit": "DX 539", "party": "Defense", "type": "Email", "admitted_trial_day": "Day 4 (April 30, 2026)", "uploaded_box_pt": "2026-04-30T15:13:27", "uploader": "Morrison Foerster", "pages": 2, "size_bytes": 150275, "source_pdf": "DX-539.pdf", "pdf_url": "https://media.mts-in.com/DX-539.pdf", "body_markdown": "# DX 539 — June 10, 2016 Musk → Birchall, \"I don't want Sam on the lease\"\n\n> The two-message Pioneer Building lease thread in which Musk tells Birchall the building should be \"viewed as a Musk Foundation building\" and \"**I don't want Sam on the lease**\" — defense's foundational receipt that Musk personally controlled the Pioneer Building from inception, not OpenAI.\n\n## Document type\n**Email thread, plain text, two messages.** Musk's June 10, 2016 7:10 AM \"Great\" reply on top of Birchall's June 10, 2016 12:06 AM email, both quoting Musk's earlier same-day note and Birchall's June 9, 2016 11:48 PM original. Subject \"Re: A few things.\" Bates 2024MUSK-0005224–5225. Highly Confidential-AEO; portions of the thread (including a separate \"Musk Home Theater Upgrade\" attachment list) are redacted in black-box overlays.\n\n## Logistics\n- **Trial admission:** Day 4 (April 30, 2026). Per [[Key Themes]]: \"DX 539 — June 2016 Pioneer Building Musk-Birchall email.\" Used by defense during Birchall direct to anchor the Pioneer Building chain and rebut plaintiffs' [[PX 103]] § 17510.6 charitable-solicitation theory.\n- **Box upload:** 2026-04-30 15:13:27 PT — Day 4 mid-afternoon batch (clustered with DX 646, DX 1156, DX 1285).\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~147 KB, 2 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `539.pdf`.\n\n## Transcribed text\n\n> **From:** \"Elon Musk\" <erm@spacex.com>\n> **To:** \"Jared Birchall\" <jbirchall@muskfoundation.org>\n> **Subject:** Re: A few things\n> **Date:** Fri, 10 Jun 2016 07:10:20 -0000\n> **Importance:** Normal\n>\n> Great\n\n> **From:** Jared Birchall <jbirchall@muskfoundation.org>\n> **To:** Elon Musk <erm@spacex.com>\n> **Subject:** Re: A few things\n> **Date:** Fri, Jun 10, 2016 at 12:06 AM\n>\n> Sounds good. Atit had Sam's name on the lease docs so I assumed that was something you two discussed. Bad assumption. I'll tell him to remove it.\n\n> **From:** Elon Musk <erm@spacex.com>\n> **To:** Jared Birchall <jbirchall@muskfoundation.org>\n> **Subject:** Re: A few things\n> **Date:** Friday, June 10, 2016\n>\n> Why are we asking for Altman's signature? Since I'm personally on the hook, this should be viewed as a Musk Foundation building, in which we will house OpenAI, Neuralink and maybe some SpaceX or Tesla people. **I don't want Sam on the lease.**\n>\n> For the LoC, what matters is how much capital I have to hold against it. If I have to reserve $400k for ten years, that is almost the same as spending $400k. I'm putting my name on the lease, so this should really be zero.\n>\n> [black-box redaction]\n\n> **From:** Jared Birchall <jbirchall@muskfoundation.org>\n> **To:** Elon Musk <erm@spacex.com>\n> **Subject:** A few things\n> **Date:** Thu, Jun 9, 2016 at 11:48 PM\n>\n> **Pioneer Building Lease** — We finalized the revisions and are awaiting Sam Altmans' signature. He's traveling in Europe, but he will get it to us by tomorrow morning. The final detail is a security deposit. Atit suggests using a letter of credit for $400k. I am awaiting a callback from City National, but I believe this will cost about $1k. The alternative is to give them $400k cash to hold. Do you have a preference?\n>\n> [black-box redaction]\n> >\n> > Jared J. Birchall\n> > Managing Director\n> > Musk Family Office\n> >\n> > <Musk Home Theater Upgrade 2016-6-7.pdf>\n> > <Musk Home Theater Upgrade Option B 2016-6-7-2.pdf>\n>\n> --\n> Jared J. Birchall\n> Managing Director\n> **Musk Family Office**\n\n## Commentary\n\nDX 539 is the cornerstone of defense's Pioneer Building counter-narrative. Plaintiffs' § 17510.6 charitable-solicitation theory ([[PX 103]]) treats the building as an OpenAI nonprofit asset for which Musk-side payments and grants in 2017–2020 were charitable solicitations within the SOL period. Defense uses DX 539 to argue that from the very first lease execution Musk himself directed the building be **\"viewed as a Musk Foundation building\"** to \"house OpenAI, Neuralink and maybe some SpaceX or Tesla people\" — i.e., a Musk-controlled multi-tenant arrangement, not an OpenAI charitable asset. The \"**I don't want Sam on the lease**\" line is the punchline: defense says no charitable solicitation of Musk by OpenAI ever existed because Musk was the landlord-equivalent. The fact that Birchall signs from `jbirchall@muskfoundation.org` while routing the lease through Atit (Choksi, Musk family office controller) reinforces the defense theme that Musk Foundation/Excession/family office were the operative actors. Cross-reference: [[PX 103]] (July 2020 Clark→Birchall reimbursement email) and the Pioneer Building sub-thread in [[Key Themes]] § \"Pioneer Building.\" Note the irrelevant home-theater attachments are visible in Birchall's signature block — a useful authenticity tell that this is a pulled-from-mailbox business email, not a constructed exhibit.\n\n---\n*See also:* [[Day 4|Day 4 digest]] · [[Pioneer Building]] · [[Jared Birchall]] · [[PX 103]] · [[Statute of Limitations]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-545", "exhibit": "DX 545", "party": "Defense", "type": "Email + attachment (lease)", "admitted_trial_day": "Day 4 (April 30, 2026)", "uploaded_box_pt": "2026-04-30T15:13:27", "uploader": "Morrison Foerster", "pages": 3, "size_bytes": 261984, "source_pdf": "DX-545.pdf", "pdf_url": "https://media.mts-in.com/DX-545.pdf", "body_markdown": "# DX 545 — June 22, 2016 YC.org → Birchall Pioneer Building \"Tenancy at Will\" lease + cover thread\n\n> Chris Clark's (YC.org) June 22, 2016 cover email to Jared Birchall (Musk Family Office) attaching the executed Tenancy-at-Will lease for OpenAI's Pioneer Building space at 3180 18th Street, San Francisco — Musk Industries LLC as tenant, OpenAI as occupant, conditioned on OpenAI's IRS Form 1023 application.\n\n## Document type\n**Email thread + attached lease contract, three pages.** Page 1: Chris Clark's June 22, 2016 5:41 AM email to Birchall (cc Nicole Imhof at YC; Sam Altman) on top of Birchall's June 17, 2016 12:53 PM email to Altman, on top of Altman's June 9, 2016 10:37 AM short email asking for OpenAI's full legal name, on top of Birchall's June 9, 2016 7:00 PM intro email. Page 2: continuation. Page 3: the **Agreement for Tenancy at Will** between Musk Industries LLC and OpenAI, Inc. (executed by OpenAI, \"Director\" signature). Bates OPENAI_MUSK00017240–17242.\n\n## Logistics\n- **Trial admission:** Day 4 (April 30, 2026) — used during Birchall direct examination. No prior wiki reference; admission day inferred from Box upload date and Day 4 mid-afternoon defense batch.\n- **Box upload:** 2026-04-30 15:13:27 PT (clustered with [[DX 502]], [[DX 1156]], [[DX 1157]]).\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~256 KB, 3 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `545.pdf`.\n\n## Transcribed text\n\n> **From:** Chris Clark <chris@ycr.org>\n> **To:** Jared Birchall <jbirchall@muskfoundation.org>\n> **Cc:** Nicole Imhof <nicole@ycombinator.com>, Sam Altman <sam@ycombinator.com>\n> **Subject:** Re: Pioneer Building Lease\n> **Date:** 6/22/2016 5:41:44 AM\n> **Attachments:** Signed_OpenAI_Pioneer_Tenancy_at_Will.pdf\n>\n> Jared,\n>\n> You should have received the Tenancy at Will agreement signed by Sam earlier this evening. I've attached a copy just in case.\n>\n> Note you will need to fill in the date of the master lease since we weren't sure if that had changed.\n>\n> Please send us a copy of the fully executed document once you have it, and thanks again for all your help!\n>\n> Chris\n\n> **From:** Jared Birchall <jbirchall@muskfoundation.org>\n> **To:** Sam Altman <sam@ycombinator.com>\n> **Cc:** Nicole Imhof <nicole@ycombinator.com>, Chris Clark <chris@ycr.org>\n> **Subject:** Re: Pioneer Building Lease\n> **Date:** Fri, Jun 17, 2016 at 12:53 PM\n>\n> Hi Sam - Not sure if you are still traveling, but attached is a 'tenancy at will' agreement between Musk Industries and OpenAI. Please sign and send back at your earliest convenience. Or if Nicole can help, that would be great as well.\n>\n> Crazier things have happened, but it is looking like we should have the lease agreement finalized today. Atit (building owner) has requested that we not spread the news until he is personally able to let a few of the other parties who are also trying to lease the building.\n>\n> Thanks!\n> Jared\n>\n> *Jared J. Birchall*\n> Managing Director\n> **Musk Family Office**\n\n> **From:** Sam Altman <sam@ycombinator.com>\n> **To:** Jared Birchall <jbirchall@muskfoundation.org>\n> **Cc:** Chris Clark <chris@ycr.org>, Nicole Imhof <nicole@ycombinator.com>\n> **Subject:** Re: Pioneer Building Lease\n> **Date:** Thu, Jun 9, 2016 at 10:37 AM\n>\n> Chris or Matt, could you please tell Jared the full legal name of the OpenAI organization?\n>\n> Nicole, Jared may need you to sign a document for me today. I'm ok with whatever he sends over.\n>\n> Thanks\n> Sam\n\n> **From:** Jared Birchall <jbirchall@muskfoundation.org>\n> **To:** Sam Altman <sam@ycombinator.com>\n> **Subject:** Pioneer Building Lease\n> **Date:** Thu, Jun 9, 2016 at 7:00 PM\n>\n> Hi Sam - I am helping Elon finalize the lease on the Pioneer building and wondered if you have a few minutes this morning to speak?\n>\n> Thank you,\n> Jared\n>\n> *Jared J. Birchall*\n> Managing Director\n> **Musk Family Office**\n\n---\n\n> **AGREEMENT FOR TENANCY AT WILL** (excerpted)\n>\n> Between **Musk Industries LLC**, a California LLC, and **OpenAI, Inc.**, a Delaware corporation.\n>\n> 1. *Recitals.*\n> (a) Musk Industries LLC is the Tenant under an Office Lease Agreement for the lease of the **Pioneer Building at 3180 18th Street, San Francisco, CA** (the \"Premises\") in which Bridgeton Pioneer Property LLC is the Landlord and Musk Industries LLC is the Tenant (the \"Pioneer Building Lease\").\n> (b) Section 25(y) of the Pioneer Building Lease provides that Musk Industries LLC may allow OpenAI to use the Pioneer Building for general office use.\n> (c) By this Agreement, Musk Industries LLC and OpenAI wish to set forth the terms of the tenancy at will allowing OpenAI permission to use and occupy the Premises.\n>\n> 2. Musk Industries LLC grants OpenAI permission to use and occupy 100% of the Premises as a tenant at will, effective on the Commencement Date, as defined in the Pioneer Building Lease.\n>\n> 3. When using the Premises, OpenAI shall comply with all of the terms and conditions of the Pioneer Building Lease.\n>\n> 4. At any time, and in its sole discretion, Musk Industries LLC may terminate OpenAI's permission to use and occupy the Premises. Musk Industries LLC shall give to OpenAI thirty (30) days written notice of such termination, in compliance with section 789 of the California Civil Code and section 1162 of the California Code of Civil Procedure.\n>\n> 5. For any period in which Musk Industries LLC allows OpenAI to use and occupy the Premises, upon ten (10) days written notice to OpenAI, Musk Industries LLC may require OpenAI to pay all of the Rent and other charges set forth in Section 4 of the Pioneer Building Lease, and all other costs and charges payable by Tenant under the Pioneer Building Lease. OpenAI shall make all such payments directly to Landlord.\n>\n> 6. **OpenAI shall not use or occupy the Premises until OpenAI has provided to Musk Industries LLC a copy of its IRS Form 1023 Application for Recognition of Exemption and all accompanying documents, as filed with the IRS.** Immediately upon receipt, OpenAI shall provide a copy of its determination letter and all correspondence from the IRS concerning OpenAI's tax exempt status and its private foundation status. **OpenAI represents and warrants that it will apply for, and expects to receive, a determination from the IRS that OpenAI will not be a private foundation.**\n>\n> Signed: **OpenAI, Inc.**, a Delaware corporation, by [Director signature]; Musk Industries LLC signature blank in attached version.\n\n## Commentary\n\nDX 545 is a quietly devastating defense exhibit. The **Pioneer Building lease** runs from **Bridgeton Pioneer Property LLC → Musk Industries LLC → OpenAI** as a tenant-at-will — meaning Musk's family-office vehicle, not Musk personally and not the Musk Foundation, was OpenAI's landlord. Defense uses this two ways: (i) it places **Jared Birchall** (Musk Family Office Managing Director) at the operational center of the Musk-OpenAI relationship as early as June 2016, supporting the [[Key Themes#Greg Brockman's contributions (\"zero\")|\"every dollar passed through Birchall\"]] cross theme on Day 4; and (ii) Section 6 — OpenAI's representation that it \"**will apply for, and expects to receive, a determination from the IRS that OpenAI will not be a private foundation**\" — is contractual evidence that all parties (Musk's family office included) **knew** OpenAI was structured as a public charity rather than a Musk-controlled private foundation, undercutting any \"Musk thought he was funding his own foundation\" theory. The 30-day termination clause at Musk Industries' \"sole discretion\" is the kind of physical-control lever defense uses to argue Musk treated OpenAI as a tenant-at-pleasure. Cross-reference: see [[DX 828]] (Birchall on the August 2018 LP term sheet — \"**plain vanilla for-profit structure**\") for Birchall's parallel skepticism of OpenAI's nonprofit framing.\n\n---\n*See also:* [[Day 4|Day 4 digest]] · [[Jared Birchall]] · [[Sam Altman]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-556", "exhibit": "DX 556", "party": "Defendants", "type": "Email thread", "admitted_trial_day": "Day 4 (April 30, 2026)", "uploaded_box_pt": "2026-04-30T15:13:27", "uploader": "Morrison Foerster", "pages": 1, "size_bytes": 373581, "source_pdf": "DX-556.pdf", "pdf_url": "https://media.mts-in.com/DX-556.pdf", "body_markdown": "# DX 556 — Aug 29–30, 2016 Musk-Altman email: \"$50 million compute donation\" from Microsoft over 3 years\n\n> Three-message thread in which Sam Altman tells Musk he has \"negotiated a $50 million compute donation\" from Microsoft, Musk says he prefers Microsoft to Amazon (\"Jeff is a bit of a tool and Satya is not\"), and Musk endorses publishing \"a simple text blog expressing appreciation of Microsoft's donation\" — the contemporaneous record of Musk personally welcoming Microsoft as OpenAI's first major compute partner.\n\n## Document type\n**Email thread, plain text, three messages.** Aug 29, 2016 10:00 AM (Altman) → 10:15 AM (Musk) → 10:18 AM (Altman) → Aug 30, 2016 6:25 AM UTC (Musk). Subject: \"Re: microsoft/openAI.\" From: Elon Musk <erm@spacex.com> to Sam Altman <sam@ycombinator.com>. Bates 2024MUSK-0005033.\n\n## Logistics\n- **Trial admission:** Day 4 (April 30, 2026) — Musk cross + redirect; Birchall direct + cross. Per the chunk wiki note: \"DX 556 — Microsoft compute donation T&C.\" Used by defense to anchor the chronology of the Microsoft relationship — Musk personally negotiated, blessed, and authored marketing-coordination terms for the **first** Microsoft compute contribution to OpenAI in August 2016, three years before the 2019 LP investment.\n- **Box upload:** 2026-04-30 15:13:27 PT — Day 4 mid-afternoon batch.\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~365 KB, 1 page.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `556 .pdf`.\n\n## Transcribed text\n\n> **From:** \"Elon Musk\" <erm@spacex.com>\n> **To:** \"Sam Altman\" <sam@ycombinator.com>\n> **Subject:** Re: microsoft/openAI\n> **Date:** Tue, 30 Aug 2016 06:25:19 -0000\n> **Importance:** Normal\n>\n> Sounds good\n>\n> I will call Satya if we get to decent terms. **A simple text blog expressing appreciation of Microsoft's donation on our website will help them a lot. They can always point people to that.**\n\n> **From:** Sam Altman <sam@ycombinator.com>\n> **To:** Elon Musk <erm@spacex.com>\n> **Subject:** Re: microsoft/openAI\n> **Date:** Mon, Aug 29, 2016 at 10:18 AM\n>\n> Amazon started really dicking us around on the T+C, especially on marketing commits.\n>\n> And their offering wasn't that good technically anyway.\n>\n> Will send you detailed terms by end of this week or so, still going back on forth on that part. So far looking clean though.\n\n> **From:** Elon Musk <erm@spacex.com>\n> **To:** Sam Altman <sam@ycombinator.com>\n> **Subject:** Re: microsoft/openAI\n> **Date:** Mon, Aug 29, 2016 at 10:15 AM\n>\n> I'm ok with this only if they don't use it in marketing. **I would also like to see the exact terms and conditions. Gifts are only as good as the T&C.**\n>\n> What happened to Amazon?\n>\n> I think Jeff is a bit of a tool and Satya is not, so I slightly prefer Microsoft, but I hate their marketing dept.\n\n> **From:** Sam Altman <sam@ycombinator.com>\n> **To:** Elon Musk <erm@spacex.com>\n> **Subject:** microsoft/openAI\n> **Date:** Mon, Aug 29, 2016 at 10:00 AM\n>\n> **I have negotiated a $50 million compute donation from them over the next 3 years!**\n>\n> Do you have any reason not to like them, or care about us switching over from Amazon?\n\n## Commentary\n\nDX 556 is the contemporaneous record of Musk personally onboarding Microsoft to OpenAI in August 2016 — three years before the 2019 limited-partnership investment that anchors plaintiffs' \"captured by Microsoft\" narrative (see [[Key Themes]] § \"'Captured by Microsoft' (statute-of-limitations fight)\" and [[PX 251]]). **Defense's read:** Musk preferred Microsoft over Amazon, said so explicitly (\"I slightly prefer Microsoft\"), reviewed the T&C personally (\"Gifts are only as good as the T&C\"), and authored the public-acknowledgment language OpenAI would use to market the gift. Eight years later he would tweet \"OpenAI is essentially captured by Microsoft\" — the defense uses this email to argue that Musk's later objection was sour grapes, not breach. **Plaintiffs' read:** the August 2016 deal was a **donation** of compute time, not a $13B equity-and-product-license restructuring; Musk's approval of a 3-year compute gift cannot be retroactively converted into approval for the 2019 capped-profit conversion or the 2023 JDCA \"all IP except AGI\" license. Note Musk's own anti-marketing instinct (\"only if they don't use it in marketing\") which becomes ironic in light of OpenAI's 2025 Microsoft IPO speculation (see [[Key Themes]] § \"The defense's 2026 valuation\"). See [[Day 4|Day 4 digest]] for the cross use.\n\n---\n*See also:* [[Day 4|Day 4 digest]] · [[PX 251]] · [[Sam Altman]] · [[Satya Nadella]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-559", "exhibit": "DX 559", "party": "Defense (OpenAI/MS)", "type": "Email thread", "admitted_trial_day": "Day 3 (April 29, 2026)", "uploaded_box_pt": "2026-04-29T15:37:43", "uploader": "Morrison Foerster", "pages": 4, "size_bytes": 420308, "source_pdf": "DX-559.pdf", "pdf_url": "https://media.mts-in.com/DX-559.pdf", "body_markdown": "# DX 559 — Dec 12, 2016 Musk \"wrong move\" / \"non-profit might... have been the wrong move\" email to Neuralink team\n\n> Musk's reply, on a Neuralink-internal NIPS-2016 recap thread from Paul Merolla, dropping the line: \"Setting it up as a non-profit might, in hindsight, have been the wrong move. Sense of urgency is not as high.\" The defense's most-on-the-nose 2016 contemporaneous Musk admission that the nonprofit structure he now calls a stolen charity had — to him — already become the *problem* a year in.\n\n## Document type\n**Email thread, plain text, two messages.** Musk's December 12, 2016 1:12 PM PST reply on top of Paul Merolla's December 12, 2016 1:06 PM email recapping NIPS 2016 (the OpenAI Universe announcement and the DeepMind multi-agent paper). All five Neuralink team addresses (Merolla, Sabes, Hanson, Hodak, Seo, Teller) on the thread. Bates 2024MUSK-0003264–0003267. Pages 2–4 are fully redacted (black boxes). Production-stamped Highly Confidential-AEO.\n\n## Logistics\n- **Trial admission:** Day 3 (April 29, 2026) — Musk direct concludes; cross begins. Per the wiki TSV: \"**DX 559 — Dec 16, 2016 Neuralink wrong-move email**\" *(note: the document is dated Dec 12, 2016, not Dec 16)*. Used by the defense on Musk's cross to puncture the \"phase one — enthusiastically supportive\" narrative for late 2016.\n- **Box upload:** 2026-04-29 15:37:43 PT — Day 3 mid-afternoon batch.\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~411 KB, 4 pages (only page 1 is substantive; pp. 2–4 are fully redacted).\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `559 .pdf`.\n\n## Transcribed text\n\n> **From:** Elon Musk <erm@neuralink.com>\n> **To:** Paul Merolla <paul@neuralink.com>\n> **Cc:** Philip Sabes <sabes@neuralink.com>, Tim Hanson <tim@neuralink.com>, Max Hodak <max@neuralink.com>, DJ Seo <dj@neuralink.com>, Sam Teller <sam@neuralink.com>\n> **Subject:** Re: Communicative rhesus?\n> **Date:** Mon, 12 Dec 2016 13:12:52 -0800\n>\n> Interesting. Deepmind is moving very fast. I am concerned that OpenAI is not on a path to catch up. Setting it up as a non-profit might, in hindsight, have been the wrong move. Sense of urgency is not as high.\n\n> **From:** Paul Merolla <paul@neuralink.com>\n> **To:** Elon Musk <erm@neuralink.com>\n> **Cc:** Philip Sabes <sabes@neuralink.com>, Tim Hanson <tim@neuralink.com>, Max Hodak <max@neuralink.com>, DJ Seo <dj@neuralink.com>, Sam Teller <sam@neuralink.com>\n> **Subject:** Communicative rhesus?\n> **Date:** Mon, Dec 12, 2016 at 1:06 PM\n>\n> Just arrived back from NIPS and wanted to share two highlights that I thought you guys might like.\n>\n> First, Ilya announced the OpenAI Universe — I am sure Sam and Elon are intimately familiar with this, but I am including for the rest of the team. Universe is an api to the digital world for AI agents, where these agents can interact with computers via pixels and keyboard / mouse commands. They already support dozens of games and physics environments, as well as an interface for internet browsing. Ilya showed and example where an agent takes an instruction such as \"Buy an airline ticket ...\" and learns to perform the appropriate sequence of actions (see the blog). Down the road, this could be a great training environment for our BCI (perhaps coupled to an AI agent). Certainly would make for a really cool demo going beyond the typical \"mind moving a mouse\"-like demos.\n>\n> Second, Nando's group at DeepMind had a paper *Learning to Communicate with Deep Multi-Agent Reinforcement Learning*. In my mind, this paper attempts to address one of the fundamental questions in AI: How can we train multiple agents with limited communication to coordinate and solve a complex task? They created an end-to-end reinforcement learning system where remarkably, the agents learn a communication protocol (language?) to solve some interesting puzzles. Although the work is preliminary, I think we could leverage this type of training to connect a neural system with an AI agent to maximize their shared utility :).\n>\n> -Paul\n\n*[Pages 2–4: fully redacted (black-box) — no transcribable content.]*\n\n## Commentary\n\nDX 559 is a defense gem because it is *Musk's own words* in *December 2016* — i.e., year one of a structure he would later (in 2024) sue to unwind as a charitable-trust breach — calling the nonprofit form \"**the wrong move**.\" Defense uses it against Musk's \"**three phases**\" narrative: he testified phase two (\"a little uncertain\") didn't begin until \"2017, 2018\" (`042926TT.txt:7148`), but here he is on December 12, 2016 already telling his Neuralink staff the nonprofit lacks \"sense of urgency.\" The exhibit also threads forward to the Karpathy-endorsed \"**OpenAI to attach to Tesla as its cash cow**\" reply (DX 749) and Musk's Nov. 2018 [[DX-844]] \"attempt that through Tesla instead\" line — see [[Key Themes]] §\"Tesla / xAI as Musk's own AGI play (the parallel).\" That Musk sent it from a *Neuralink* email, to *Neuralink* staff, also helps the defense's larger Musk-empire frame: by late 2016 he was already mentally ranking OpenAI against Musk-controlled vehicles and finding it wanting.\n\n---\n*See also:* [[Day 3|Day 3 digest]] · [[DX-844]] · [[DX-758]] · [[DX-853]] · [[Key Themes]] · [[Quotes]]\n"}
{"exhibit_id": "DX-600", "exhibit": "DX 600", "party": "Defense", "type": "Email thread", "admitted_trial_day": "Day 4 (April 30, 2026)", "uploaded_box_pt": "2026-04-30T15:13:31", "uploader": "Morrison Foerster", "pages": 1, "size_bytes": 728950, "source_pdf": "DX-600.pdf", "pdf_url": "https://media.mts-in.com/DX-600.pdf", "body_markdown": "# DX 600 — Jan 18, 2017 Birchall → Sam Teller, \"Elon holds 100 percent of the power and authority regarding the building\"\n\n> Birchall's two-sentence answer to Tesla's chief of staff confirming that **Elon Musk personally — not OpenAI — holds the controlling lease interest in the Pioneer Building**, and that Musk had earmarked the upper floors for **Neuralink**, not OpenAI.\n\n## Document type\n**Email thread, plain text, three messages.** Sam Teller (steller@tesla.com) → Jared Birchall (jbirchall@muskfoundation.org), Jan 18, 2017 4:29 PM, \"It's Elon on the lease, right?\"; Birchall reply 4:38 PM; Teller's \"Thanks. I'll try to get to the bottom of this on Tuesday.\" reply at 5:40:04 PM. Subject: \"Re: Pioneer.\" Bates TESLA_000000897. Marked HIGHLY CONFIDENTIAL — AEO.\n\n## Logistics\n- **Trial admission:** Day 4 (April 30, 2026) — Birchall direct + cross. Used by defense alongside [[DX 539]] in the [[Key Themes]] §\"Pioneer Building\" sequence to anchor that the master lessee was Musk Industries LLC (see [[PX 79]]) and that Musk personally controlled the building.\n- **Box upload:** 2026-04-30 15:13:31 PT — Day 4 mid-afternoon Defense batch.\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~712 KB, 1 page.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `600.pdf`.\n\n## Transcribed text\n\n> **From:** Sam Teller <steller@tesla.com>\n> **To:** Jared Birchall <jbirchall@muskfoundation.org>\n> **Subject:** Re: Pioneer\n> **Date:** 1/18/2017 5:40:04 PM\n>\n> Thanks. I'll try to get to the bottom of this on Tuesday.\n\n> **From:** Jared Birchall <jbirchall@muskfoundation.org>\n> **To:** Sam Teller <steller@tesla.com>\n> **Subject:** Re: Pioneer\n> **Date:** Wednesday, January 18, 2017 at 4:38 PM\n>\n> Yes, **Elon holds 100% of the power and authority regarding the building**. When OpenAI first moved in he made it clear that he wanted them to stay only on the first floor. He was earmarking the other spaces for Neuralink. It's possible Greg had a sidebar conversation with him about using other spaces. He may have casually approved not knowing they were planning on moving forward.\n\n> **From:** Sam Teller <steller@tesla.com>\n> **To:** Jared Birchall <jbirchall@muskfoundation.org>\n> **Subject:** Pioneer\n> **Date:** Wed, Jan 18, 2017 at 4:29 PM\n>\n> It's Elon on the lease, right?\n\n## Commentary\n\nDX 600 is defense's six-line confirmation of the [[Key Themes]] §\"Pioneer Building\" structural fact: **Musk personally — not OpenAI — controlled the building** that defense will later argue was leased through Musk Industries LLC for Musk's own purposes (Neuralink space-banking on the upper floors), with OpenAI's occupancy as a first-floor sub-permitted use. The Birchall-Teller exchange is also useful for defense to establish that Musk's \"**I don't want Sam on the lease**\" instruction in [[DX 539]] (June 2016) was not a one-time decision but a continuing policy that Birchall and Teller were administering through 2017. Plaintiffs counter with [[PX 60]] (the Vanguard Charitable grants paying OpenAI's monthly rent in 2017) and [[PX 103]] (the July 2020 Clark email asserting the for-profit OpenAI was \"happy to cover 100 percent of the rent\") — the predicates for plaintiffs' alternative § 17510.6 charitable-solicitation theory. Pair with [[PX 79]] (the master lease itself) to read DX 600 against the actual Musk Industries LLC tenant rep and Musk personal guaranty.\n\n---\n*See also:* [[Day 4|Day 4 digest]] · [[PX 79]] · [[PX 60]] · [[PX 103]] · [[DX 539]] · [[Jared Birchall]] · [[Sam Teller]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-619", "exhibit": "DX 619", "party": "Defense", "type": "Text messages", "admitted_trial_day": "Day 8 (May 6, 2026)", "uploaded_box_pt": "2026-05-06T14:57:03", "uploader": "Morrison Foerster", "pages": 2, "size_bytes": 530174, "source_pdf": "DX-619.pdf", "pdf_url": "https://media.mts-in.com/DX-619.pdf", "body_markdown": "# DX 619 — June 9–20, 2017 Sam Teller / Shivon Zilis / Emma Gallagher texts re: poaching Andrej Karpathy from OpenAI\n\n> A two-page text-thread screenshot from June 9–20, 2017 — Tesla/Musk staff (Sam Teller, Shivon Zilis, Emma Gallagher) coordinating a Tesla offer for Andrej Karpathy while Karpathy was still at OpenAI. \"**Signed offer in hand for Andrej!**\" — \"Fuck yeahhhhhhh.\" — \"When does E plan to announce to team? Any sense of how mad OpenAI will be? :)\"\n\n## Document type\n**Text messages, screenshot from iMessage.** Two pages of conversation bubbles. Three participants: Sam Teller, Shivon Zilis (\"S\"), Emma Gallagher. Date range June 9, 2017 4:44 PM – June 20, 2017 11:24 AM. Bates GALLAGHER00000165–00000166. Marked Highly Confidential.\n\n## Logistics\n- **Trial admission:** Day 8 (May 6, 2026 — Murati live; Zilis live; Toner video begins). No prior wiki citation; admission day inferred from Box upload and Zilis's live testimony that day. Likely used in connection with [[Key Themes|\"Tesla / xAI as Musk's own AGI play\"]] — \"Karpathy compiled a poach list for Musk **while Musk was still on OpenAI's board**\" *(5/6/2026 Testimony @ ~13:02 PT)*.\n- **Box upload:** 2026-05-06 14:57:03 PT — Day 8 mid-afternoon defense batch (clustered with DX 664, DX 724, DX 757, DX 824, DX 900).\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~518 KB, 2 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `0619.pdf`.\n\n## Transcribed text\n\n> [Page 1 — conversation between Sam Teller, Shivon Zilis (\"S\"), Emma Gallagher.]\n>\n> **Emma Gallagher:** Milan just cruised over to talk – I'll report back\n>\n> **Shivon:** I can't image he wouldn't accept\n>\n> **Shivon:** I just don't know if that has implications to OpenAI call timing\n>\n> **Jun 9, 2017 at 4:44 PM**\n>\n> **Emma Gallagher:** Signed offer in hand for Andrej!\n>\n> **Shivon:** Fuck yeahhhhhhh\n>\n> **Emma Gallagher:** He was like \"oh, sorry didn't realize I was holding anything up. I'll send over now\" :)\n>\n> **Sam Teller:** Haha\n>\n> **Shivon:** Oh, he seems so happy :)\n>\n> **Sam Teller:** When does E plan to announce to team? Any sense of how mad OpenAI will be? :)\n>\n> [Page 2 — continuation.]\n>\n> **Sam Teller:** When does E plan to announce to team? Any sense of how mad OpenAI will be? :)\n>\n> **Shivon:** Good question on OpenAI. Talked to Greg today and he clearly had no idea. Andrej mentioned he's talked to a few people about it at OpenAI for friend advice but they have clearly kept their mouths shut so far\n>\n> **Shivon:** Tbd on timing\n>\n> **Jun 20, 2017 at 11:24 AM**\n>\n> **Sam Teller:** Autopilot regression\n>\n> [Sam Teller]: He is going to fire Lattner this morning\n>\n> **Emma Gallagher:** Fuck\n>\n> **Emma Gallagher:** Who?\n>\n> **Sam Teller:** Elon Musk, CEO of Tesla\n>\n> **Emma Gallagher:** Haha I thought you said resignation\n>\n> **Emma Gallagher:** Are we going to start with Andrej\n\n## Commentary\n\nDX 619 is one of defense's strongest **\"parallel AGI play\"** exhibits — but it cuts both ways and that is exactly why both sides are interested in it. **Defense's read:** Andrej Karpathy's June 2017 move from OpenAI to Tesla was a normal hiring event coordinated through Musk's office — the surrounding Day 8 framing (see [[Key Themes]] § \"Bury this in Tesla for stealth advantage\") shows Tesla was a *competing* AGI venture all along, undermining Musk's \"I gave up on OpenAI only after they betrayed me\" narrative. **Plaintiffs' read:** the same texts show Musk's chief-of-staff Sam Teller and Shivon Zilis (who sat on OpenAI's board the next year) actively *coordinating* the poach in June 2017 — when **Musk was still on OpenAI's board** — and stage-managing the disclosure (\"**Talked to Greg today and he clearly had no idea**\"). Schubert on Day 8 made the \"while Musk was still on OpenAI's board\" point explicit *(5/6/2026 Testimony @ ~13:02 PT)*. The Chris Lattner cameo (\"He is going to fire Lattner this morning … Elon Musk, CEO of Tesla\") is incidental Tesla-Autopilot color — Lattner left Tesla a few weeks before Karpathy arrived. Cross-reference the **DX 758 \"OpenAI possible scenarios\"** memo (scenario 4: \"OpenAI becomes a B Corp subsidiary of Tesla\") and Zilis's \"**bury this in Tesla for stealth advantage**\" text (Feb 2018) — see [[Key Themes]] § \"Day 8 — what Zilis added.\"\n\n---\n*See also:* [[Day 8|Day 8 digest]] · [[Shivon Zilis]] · [[Andrej Karpathy]] · [[Sam Teller]] · [[DX-757]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-621", "exhibit": "DX 621", "party": "Defense", "type": "Charitable application form", "admitted_trial_day": "Day 4 (April 30, 2026)", "uploaded_box_pt": "2026-04-30T15:13:31", "uploader": "Morrison Foerster", "pages": 10, "size_bytes": 1203598, "source_pdf": "DX-621.pdf", "pdf_url": "https://media.mts-in.com/DX-621.pdf", "body_markdown": "# DX 621 — June 2017 Musk Foundation Fidelity Charitable DAF application\n\n> Elon Musk's signed June 21, 2017 Fidelity Charitable \"Charitable Investment Advisor Program\" application establishing **The Musk Foundation Charitable Fund** donor-advised fund — used by defense to make the donor-advised-fund / tax-deduction point on Birchall cross.\n\n## Document type\n**Fidelity Charitable Charitable Investment Advisor Program (CIAP) application form, 10 pages.** Pre-printed Fidelity form completed in handwriting and signed; production stamps \"HIGHLY CONFIDENTIAL\" and Bates FIDCHAR-OPENAI-000751–000760. Document ID DX-0621. Court-stamp lower right indicates \"Case No. 4:24-CV-04722-YGR / Musk et al. v. Altman et al. / Exhibit No. DX-0621.\"\n\n## Logistics\n- **Trial admission:** Day 4 (April 30, 2026) — Musk cross + redirect; Birchall direct + cross. Pertinent to the **donor-advised funds / tax-deduction wrinkle** ([[Key Themes]]) — Birchall on cross conceded \"I'm not familiar with all of the legalities of donor advised funds\" and could not say whether Musk could ever recover money once parked in a DAF.\n- **Box upload:** 2026-04-30 15:13:31 PT — Day 4 mid-afternoon batch (clustered with DX 668 and DX 1256).\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~1.2 MB, 10 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `621.pdf`.\n\n## Transcribed text\n\n> **FIDELITY Charitable — Charitable Investment Advisor Program Application**\n>\n> *This Program allows Account Holders with more than $250,000 in a Giving Account® to nominate their investment advisor to manage Fidelity Charitable's assets.*\n>\n> **1. Account Holder Information**\n>\n> [B.] **New Giving Account Holder** — checked.\n>\n> **Primary Account Holder:**\n> First Name: **Elon**\n> Last Name: **Musk**\n> SSN/TIN: [redacted]\n> Date of Birth / Email / Phone: [redacted]\n> Citizenship: **U.S. citizen**\n>\n> **Name Your Giving Account:** *The Musk Foundation Charitable Fund*\n>\n> **Mailing/Legal address:** [redacted, USA]\n>\n> **Successor Election — Individual Successor:** 100% of Giving Account\n> First Name: **Kimbal**\n> Last Name: **Musk**\n> Relationship to Account Holder: **Brother**\n> Citizenship: U.S. citizen\n>\n> **2. Nominated Advisor Firm Information**\n>\n> Firm Name: **UBS FINANCIAL SERVICES**\n> Firm Mailing Address: 2000 AVENUE OF THE STARS, 7TH FL NORTH, LOS ANGELES, CA 90067, USA\n> Primary Advisor Representative Name: **M. SCOTT HARRIES**\n> Advisor Phone: 310-734-2414\n> Advisor Email: **SCOTT.HARRIES@UBS.COM**\n> Birthday: 02/04\n>\n> **C. Investment Advisory Fee:** Basis Points: **0**\n>\n> **D. Benchmark Selection:**\n> Barclays 3-Month U.S. Treasury Bill Index — **100%**\n> Total: 100%\n>\n> **E. Dividend and Capital Gain Payments:** **Pay All to Core Account** (checked).\n>\n> **3. Irrevocable Contribution**\n>\n> A. Where Will the Advisor Firm Custody the Charitable Assets?\n> ☒ **Other** — Firm Name: **UBS FINANCIAL SERVICES**\n>\n> *(B. What Assets Are You Contributing — left blank on this version of the form; cash/securities sections unfilled, indicating the funding instructions were issued via separate Letter of Instruction and/or wire pursuant to UBS custody.)*\n>\n> **4. Signatures and Dates**\n>\n> Primary Account Holder Name: **ELON MUSK**\n> Primary Account Holder Signature: */s/ Elon Musk/*\n> Date: **06/21/2017**\n>\n> Investment Advisor Name: **M. SCOTT HARRIES**\n> Investment Advisor Signature: */s/ M. Scott Harries/*\n> Date: **6/22/17**\n>\n> *(Section 5, Additional Account Holders — left blank. Section 6, Terms and Conditions — pre-printed boilerplate governing Fidelity Brokerage authorizations, agency, indemnification, Massachusetts choice-of-law. Page 10 — pre-printed index disclosures and Fidelity Charitable trademark notice.)*\n>\n> **Acknowledgments (Section 4):**\n> \"Understands that any contribution, once accepted by the Trustees, represents an **irrevocable contribution to Fidelity Charitable and is not refundable**.\"\n\n## Commentary\n\nDX 621 is the founding instrument of the **Fidelity Charitable donor-advised fund** that Musk used to route a substantial portion of the contributions plaintiffs claim were \"stolen\" by OpenAI. The exhibit's force is procedural rather than substantive: the pre-printed acknowledgment that the contribution is \"**irrevocable**\" and \"**not refundable**\" — combined with Musk's signature — is the predicate for defense's [[Key Themes#The \"donor-advised funds\" / tax-deduction wrinkle|donor-advised-fund / tax-deduction line of cross]] (\"for every single one of those donations, it's somewhere in the chain, Mr. Musk was entitled to take a tax deduction\" — Wilson to Birchall, `043026TT.txt:5413`). Once contributed to a DAF, Musk cannot personally recover the funds — he can only recommend grants — which complicates plaintiffs' restitution-to-Musk theory the judge previewed on Day 5 (\"**it goes back to the plaintiff**\" — `050126TT.txt:1446`). The June 2017 timing also pre-dates the August–September 2017 Brockman/Sutskever \"Honest Thoughts\" ([[PX 157]]) negotiation by a few weeks, fitting defense's narrative that Musk's giving infrastructure was being industrialized in parallel with — not in reliance on — the OpenAI nonprofit promise.\n\n---\n*See also:* [[Day 4|Day 4 digest]] · [[Jared Birchall]] · [[Musk Foundation]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-622", "exhibit": "DX 622", "party": "Defense", "type": "Email thread", "admitted_trial_day": "Day 7 (May 5, 2026)", "uploaded_box_pt": "2026-05-05T14:27:15", "uploader": "Morrison Foerster", "pages": 2, "size_bytes": 545873, "source_pdf": "DX-622.pdf", "pdf_url": "https://media.mts-in.com/DX-622.pdf", "body_markdown": "# DX-622 — July 1–3, 2017 OpenAI biweekly update: Cerebras, AGI in 5 years, and the Tesla-chip aside\n\n> Brockman's biweekly leadership update declaring OpenAI sees \"a concrete path to AGI within 5 years,\" with Musk replying that the Tesla chip team might \"exceed the reality of what Cerebras claims.\"\n\n## Document type\n**Email thread, plain text, multiple replies.** Biweekly project-status email from Greg Brockman to the OpenAI leadership distribution, with replies from Sam Altman, Ilya Sutskever, and Elon Musk between July 1 and July 3, 2017. Bates 2024MUSK-0009473–0009474.\n\n## Logistics\n- **Trial admission:** Day 7 (May 5, 2026) — Brockman defense direct + Mr. Molo recross; Wu deposition.\n- **Box upload:** 2026-05-05 14:27:15 PT — same-day Day-7 uploader batch.\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~546 KB, 2 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `0622.pdf`.\n\n## Transcribed text\n\n> **From:** Elon Musk <erm@spacex.com>\n> **To:** Ilya Sutskever <ilyasu@openai.com>\n> **Cc:** Greg Brockman <gdb@openai.com>, Sam Altman <shga@ycombinator.com>, Elon Musk <erm@openai.com>\n> **Subject:** Re: openai biweekly update\n> **Date:** Mon, 3 Jul 2017 17:19:42 +0000\n>\n> As mentioned, Jim Keller (who is the real deal) says those guys always oversell their abilities. Before we make any big decisions, I'd like to hear what the Tesla chip team can do. Might be able to exceed the reality of what Cerebras claims to be making.\n\n> **From:** Ilya Sutskever <ilyasu@openai.com>\n> **To:** Sam Altman <shga@ycombinator.com>\n> **Cc:** Greg Brockman <gdb@openai.com>, Elon Musk <erm@openai.com>\n> **Subject:** Re: openai biweekly update\n> **Date:** Sat, Jul 1, 2017 at 3:23 PM\n>\n> They do not, at least not yet.\n\n> **From:** Sam Altman <shga@ycombinator.com>\n> **To:** Greg Brockman <gdb@openai.com>\n> **Cc:** Ilya Sutskever <ilyasu@openai.com>, Elon Musk <erm@openai.com>\n> **Subject:** Re: openai biweekly update\n> **Date:** Sat, Jul 1, 2017 at 3:19 PM\n>\n> Does Cerebras believe AGI is 5 years away with their product?\n\n> **From:** Greg Brockman <gdb@openai.com>\n> **To:** Sam Altman <shga@ycombinator.com>\n> **Cc:** Ilya Sutskever <ilyasu@openai.com>, Elon Musk <erm@openai.com>\n> **Subject:** Re: openai biweekly update\n> **Date:** Sat, 1 July 2017 at 15:16\n>\n> Very slim odds of winning that way. They're 1.5 years in (plus experience trying this at a previous startup). The Cerebras hardware is a breakthrough and we've obtained the details necessary for our assessment under NDA. The other deep learning hardware startups (Graphcore, Groq) are not building anything as ambitious.\n>\n> We should assume that Google is already working on TPU 3.0 (they were working on TPU 2.0 before 1.0 was publicly announced), which is likely to be a generation behind gen 1 of Ceberas. So we estimate that Cerebras is somewhere between 2-3 years ahead of Google, leaving a narrow window of opportunity.\n>\n> We started OpenAI without any special advantage over our competitors — except for the mission. We spent a year learning how to operate and the past 6 months ramping up our execution. Joining forces with Cerebras gives us an unfair advantage over the competition, but starting our own from-scratch effort leaves us again a step behind Google.\n>\n> - gdb\n\n> **From:** Sam Altman <shga@ycombinator.com>\n> **To:** Greg Brockman <gdb@openai.com>\n> **Cc:** Ilya Sutskever <ilyasu@openai.com>, Elon Musk <erm@openai.com>\n> **Subject:** Re: openai biweekly update\n> **Date:** Sat, Jul 1, 2017 at 2:32 PM\n>\n> If we cannot acquire or exclusively partner with Cerebras, can we help start/fund another company to make a similar piece of hardware?\n\n> **From:** Greg Brockman <gdb@openai.com>\n> **To:** Sam Altman <shga@ycombinator.com>, Ilya Sutskever <ilyasu@openai.com>, Elon Musk <erm@openai.com>\n> **Subject:** openai biweekly update\n> **Date:** Sat, 1 July 2017 at 10:27\n>\n> **tl;dr: We see a path to building AGI in 5 years, several years before everyone else. This will require acquiring or otherwise merging with Cerebras, and for Cerebras to ship no more than 6-12 months behind schedule. If Google instead acquires Cerebras and Cerebras ships, then success is impossible.**\n>\n> Until recently, we believed that full AGI was 10-15 years in the future. That is, somewhere on the horizon but not imminent.\n>\n> Our beliefs have changed. **OpenAI sees a concrete path to AGI within 5 years.** We have good reason to believe that DeepMind thinks that AGI is ~10 years away. Thus, at least for a while, we have a temporary information advantage over Google, DeepMind, and everyone else. We must make the absolute best use of this information advantage.\n>\n> A few months ago, we realized we could build AGI within 2 years after gaining access to a thousand machines with 1 exaflop and 100 TB memory. A machine of such speed can run a neural net as large as a human brain for several years of thinking in one day. Such a machine has the property that every overnight experiment has a credible chance of producing AGI in the morning. We plan to use this machine in order to run a society of agents developing in a competitive environment. Agents in such an environment will learn to cooperate, communicate, develop an extremely sophisticated language, a theory of mind and excellent social skills, a market economy, a legal system, and grow in general intelligence. In short, they'll develop the cognitive toolkit of the early Sapiens — which we know is sufficient to match humans' ability on every cognitive axis. (This plan is remarkably similar to a Greg Egan short story: http://ttapress.com/553/crystal-nights-by-greg-egan/.)\n>\n> We recently visited Cerebras, saw their gen 1 hardware, and confirmed that they have a credible path for building an exaflop machine by 2020 using essentially today's technology. Scott Gray and Paul Merolla spent a day digging into their tech and believe that they will deliver with high probability. Their gen 1 hardware is going to be 1000 times faster than the best available GPU while being a lot more flexible overall (though more difficult to program) for the kind of computations that we need in AI. For a tangible example, Cerebras's gen 1 hardware will be able to train a VGG model (the largest visual recognition model, which takes days to a week on an 8 GPU machine) in 3 minutes.\n>\n> Cerebras is categorically different from other hardware projects. In particular, the TPU 2.0 pales in comparison.\n>\n> **Exclusive access to Cerebras hardware would give OpenAI an overwhelming hardware advantage over Google. By merging, we would create an entity which can build AGI in 5 years, while being several years ahead of others.** We acknowledge that things may not go as planned; however, assuming Cerebras executes their hardware reasonably on schedule, then the 5 years timeline has a 50% chance of success, and 7 years has an 80% chance of success. There is a small probability mass on AGI being dramatically harder than we imagine, in which case no one will be there within the next 20 years.\n>\n> **Lastly, if Google acquires Cerebras, there is no path for long term success for OpenAI.**\n>\n> And now, to a quick summary of our progress over the past few weeks.\n>\n> - **Robotics:** good step towards solving the Rubik's Cube\n> - **Multiagent self play:** Progress on competitive self play\n> - We solved adversarial examples, a fundamental problem in machine learning…\n> - **Dota 2:** 80% winrate against the best bot from two weeks ago\n\n## Commentary\n\nDX 622 is a defense exhibit that lands at the intersection of two trial threads. First, it bolsters the defense's \"Tesla as Musk's parallel AGI play\" theme: Musk's reflexive answer to OpenAI's hardware bottleneck is to redirect the analysis to \"the Tesla chip team\" — a 2017 datapoint for the [[Key Themes]] line that Musk wanted OpenAI's compute path to flow through Tesla. Second, it is the contemporaneous predicate for the **Cerebras conflict-of-interest** subplot ([[Greg Brockman]]'s March 17, 2017 Cerebras stock purchase, never disclosed): Brockman is here pitching the OpenAI board on a Cerebras merger while personally holding Cerebras equity. Used by [[Brockman]]'s defense direct on [[Day 7|Day 7]] to anchor the \"AGI in 5 years\" optimism that informed the August–September 2017 negotiations dramatized by [[PX 157]] and [[PX 151]] — see also the Cerebras concern flagged by Sutskever/Brockman themselves in the \"Honest Thoughts\" letter.\n\n---\n\n*See also:* [[Day 7|Day 7 digest]] · [[Greg Brockman]] · [[Brockman Journal]] · [[PX 157]] · [[PX 151]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-626", "exhibit": "DX 626", "party": "Defense", "type": "Text messages", "admitted_trial_day": "Day 7 (May 5, 2026)", "uploaded_box_pt": "2026-05-05T14:26:59", "uploader": "Morrison Foerster", "pages": 3, "size_bytes": 204727, "source_pdf": "DX-626.pdf", "pdf_url": "https://media.mts-in.com/DX-626.pdf", "body_markdown": "# DX 626 — July 14, 2017 Brockman/Zilis texts (Dota, Cerebras, Gates donation)\n\n> A 27-message single-day text thread between Greg Brockman and Shivon Zilis on July 14, 2017 — covering The International Dota announce plan, GPU asks (\"said do what we need to do\"), and Brockman's note that Altman is **\"going to sun valley to ask gates to donate\"** and that nonprofit \"may not be the right one now.\"\n\n## Document type\n**Text messages, Cellebrite-style \"Short Message Report\" extraction.** Three pages: cover sheet (1 conversation `677f8110c60b1e12b34804557f6829fe`, 27 messages, 2 participants — Greg Brockman, Shivon Zilis) plus two pages of message-by-message transcript with timestamps in GMT+00:00. Bates ZILIS-0001517–1519. Document ID DX-0626.\n\n## Logistics\n- **Trial admission:** Day 7 (May 5, 2026) — used during the Brockman defense direct and Mr. Molo's recross block to anchor the July 2017 timing of the Cerebras \"reverse merger\" idea, the structure debate, and the Gates donation outreach. No prior wiki reference; admission day inferred from Box upload date.\n- **Box upload:** 2026-05-05 14:26:59 PT — Day 7 mid-afternoon batch (clustered with DX 748).\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~200 KB, 3 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `0626.pdf`.\n\n## Transcribed text\n\n> **Conversation:** 677f8110c60b1e12b34804557f6829fe\n> **Date Range:** 7/14/2017\n> **Participants:** Greg Brockman, Shivon Zilis\n>\n> **Messages in chronological order (GMT +00:00):**\n>\n> **SZ** (Shivon Zilis) — 7/14/2017, 2:15 AM\n> Heads up I will try to join but likely in flight\n>\n> **GB** (Greg Brockman) — 7/14/2017, 2:15 AM\n> cool\n>\n> **SZ** — 7/14/2017, 5:35 AM\n> How did it go?\n>\n> **GB** — 7/14/2017, 5:35 AM\n> Went well!\n>\n> **GB** — 7/14/2017, 5:37 AM\n> - ocean: agreed on announcing around the international; he suggested playing against the best player from the winning team which seems cool to me. i asked him to call gabe from valve and he said he would. I think this is better than our default of announcing in advance we've beaten the best 1v1 player and then having our bot playable at a terminal at TI (valve's offered to give us terminals for people to play).\n>\n> **GB** — 7/14/2017, 5:37 AM\n> - gpus: said do what we need to do\n>\n> **GB** — 7/14/2017, 5:41 AM\n> - cerebras: we talked about the reverse merger idea a bit. independent of cerebras, turned into talking about structure (he said non-profit was def the right one early on, may not be the right one now — ilya and I agree with this for a number of reasons). He said he's going to sun valley to ask gates to donate.\n>\n> **SZ** — 7/14/2017, 5:43 AM\n> Gates and others. Will try to work it for ya\n>\n> **SZ** — 7/14/2017, 5:43 AM\n> Do you know if Emma was on?\n>\n> **GB** — 7/14/2017, 5:43 AM\n> Pretty sure yes\n>\n> **GB** — 7/14/2017, 5:43 AM\n> He dropped at one point and she was on then\n>\n> **SZ** — 7/14/2017, 5:43 AM\n> I will remind him on Valve tomorrow when I see him if not\n>\n> **SZ** — 7/14/2017, 5:43 AM\n> Ah ok\n>\n> **GB** — 7/14/2017, 5:44 AM\n> Btw you should still join slack!\n>\n> **SZ** — 7/14/2017, 5:44 AM\n> Oh geez yes I need to. Would be such good learning and context!!\n>\n> **GB** — 7/14/2017, 5:44 AM\n> currently talking about the 1v1 announce plans there, would be great to have you in the convos\n>\n> **SZ** — 7/14/2017, 5:46 AM\n> In! Thank you for nudge!\n>\n> **SZ** — 7/14/2017, 3:19 PM\n> I'm confused as to how much 1v1 vs 5v5 matters\n>\n> **SZ** — 7/14/2017, 3:19 PM\n> Technical feats aside\n>\n> **SZ** — 7/14/2017, 3:21 PM\n> I mean just in the world. TI is focused on 5v5 teams, not individuals? Or do both get reasonable prominence\n>\n> **GB** — 7/14/2017, 4:37 PM\n> TI used to have a 1v1 event\n>\n> **GB** — 7/14/2017, 4:37 PM\n> But not this year\n>\n> **GB** — 7/14/2017, 4:37 PM\n> Some tournaments still do it\n>\n> **GB** — 7/14/2017, 5:01 PM\n> Our expert says it's a side event, but cool to see because you don't see it all the time. Cool to see one person facing off against another directly\n>\n> **SZ** — 7/14/2017, 7:46 PM\n> Went well with Gates\n>\n> **SZ** — 7/14/2017, 7:47 PM\n> Will be a follow up that they will communicate\n>\n> **GB** — 7/14/2017, 7:53 PM\n> Awesome!\n>\n> [end of conversation]\n\n## Commentary\n\nDX 626 is a small but high-density defense exhibit. Three things land at once on July 14, 2017: (i) Brockman reports a meeting in which Musk (\"he\") said GPUs are go (\"**said do what we need to do**\") and that **the non-profit \"may not be the right one now\"** — i.e., Musk himself, six weeks before the August 2017 control fight in [[PX 151]], was already contemplating moving off the 501(c)(3); (ii) Musk was personally going to Sun Valley \"to ask Gates to donate,\" which the defense uses to argue that the sole identifiable charitable solicitation of Bill Gates was a *Musk* solicitation, not an OpenAI-of-Musk solicitation — relevant to plaintiffs' § 17510.6 theory; (iii) Brockman is already raising the **Cerebras \"reverse merger\" idea** with Musk — defense uses this to neutralize the plaintiffs' Cerebras-conflict-of-interest narrative built on Brockman's earlier Cerebras stock purchase by showing Musk was an active participant in those discussions. The Dota-1v1-vs-5v5 thread is largely color but it dates the OpenAI Five competitive-AI program ([[PX 233]] picks up the same thread in March/April 2018). Cross-reference: Sutskever's \"non-profit was successful because both you and Sam were in it\" line in [[PX 157]] reads against this earlier \"may not be the right one now\" message.\n\n---\n*See also:* [[Day 7|Day 7 digest]] · [[Greg Brockman]] · [[Shivon Zilis]] · [[PX 157]] · [[PX 151]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-627", "exhibit": "DX 627", "party": "Defense", "type": "Email", "admitted_trial_day": "Day 8 (May 6, 2026)", "uploaded_box_pt": "2026-05-06T14:57:04", "uploader": "Morrison Foerster", "pages": 1, "size_bytes": 391304, "source_pdf": "DX-627.pdf", "pdf_url": "https://media.mts-in.com/DX-627.pdf", "body_markdown": "# DX 627 — July 18, 2017 Zilis → Musk pre-meeting \"OpenAI notes\" (mission, hardware, Cerebras, Dota)\n\n> Shivon Zilis's July 18, 2017 briefing email to Elon Musk and Sam Teller previewing that night's conversation with Greg Brockman and Ilya Sutskever — sketching OpenAI's mission shift toward \"**build AGI before anyone else**,\" the hardware-as-critical-factor framing, the Cerebras-vs-billion-cores choice (with the parenthetical \"**assuming OpenAI converts from being a non-profit, would be more valuable??**\"), and the $50M Dota 5v5 compute ask.\n\n## Document type\n**Email, plain text, single message.** Shivon Zilis (then `shivon@openai.com`) → Elon Musk; Sam Teller. Subject \"OpenAI notes in advance of tonight's meeting.\" Date Tuesday, July 18, 2017 1:24:27 PM. No replies. Bates not visible (Zilis production). Document ID DX-0627.\n\n## Logistics\n- **Trial admission:** Day 8 (May 6, 2026). Pre-trial wiki reference: **\"DX 627 — July 2017 Zilis email re: Cerebras + conversion.\"** Used during the Zilis live direct examination on Day 8.\n- **Box upload:** 2026-05-06 14:57:04 PT — Day 8 mid-afternoon defense batch (clustered with [[DX 651]], [[DX 712]], [[DX 1017]], [[DX 1290]]).\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~382 KB, 1 page.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `0627.pdf`.\n\n## Transcribed text\n\n> **From:** Shivon Zilis\n> **To:** Elon Musk; Sam Teller\n> **Subject:** OpenAI notes in advance of tonight's meeting\n> **Date:** Tuesday, July 18, 2017 1:24:27 PM\n>\n> A quick heads up what the conversation with Greg and Ilya will look like:\n>\n> **OpenAI's Mission:** Want to confirm what it should be. Given the AGI timelines keeps compressing, they think it should be \"build AGI before anyone else\" (thinking of Google and China, specifically). This isn't a departure, per se, but want to confirm that you still think the best possible future would be brought about this way? This would of course involve being much less \"open\" and require decisions that would prioritize speed over other considerations (safety, government involvement, etc).\n>\n> **Hardware:** if it's true that OpenAI should dead set on building AGI before anyone else, the critical factor is hardware. They think if all resources exhausted, best you can get in terms of human capital / ingenuity over DeepMind 2-3x, whereas if you buy that HW is really going to depart from Moore's law and hit 10x per year it's not even a contest as to which battlefield more critical right now. They are at the point where they want to bet the farm on the right hardware strategy.\n>\n> **Cerebras v A Billion Cores:** In an ideal world both would be pursued. Cerebras is a high beta experiment, they understand that, but think that if it succeeds it's so game changing that if Google gets it, it's game over re: race for AGI. Reason is that even if we started on a billion core system today, it's still 18 months behind and that's enough to lose the battle. Want to keep working on billion core idea since higher probability of eventual success but think it's prudent to figure out the right strategy to still merge or exclusively partner with Cerebras as well. They believe the combined entity, **assuming OpenAI converts from being a non-profit, would be more valuable??** (Note: Andrew Feldman reached out to them to discuss next steps. That meeting is planned for next week)\n>\n> **Dota supremacy... but at what cost:** The compute for Dota 5v5 is hard to estimate but would get into the $50M range on the high side. They don't have perfect data but estimate that AlphaGo was somewhere in this ballpark. That begs the question of is this worth the price tag? They think yes, since it's likely to have an equal or better outcome to AlphaGo but want to both confirm that you agree and also make sure there is a path to funding should it come to that.\n\n## Commentary\n\nDX 627 is a Day-8 Zilis-direct cornerstone for the defense's [[Key Themes#\"Unequivocal control\" (the 2017 negotiation breakdown)|2017 negotiation backstory]]. Read in context, three lines do real work: (i) the **mission reframe** — \"**build AGI before anyone else**\" — is the founders telling Musk *to his face* in summer 2017 that \"**being much less 'open'**\" was on the table; defense uses this to argue Musk acquiesced in the openness retreat that he later sued over; (ii) the parenthetical \"**assuming OpenAI converts from being a non-profit, would be more valuable??**\" is the earliest known **pre-PX-157** evidence that the *founders themselves* were openly discussing for-profit conversion in 2017 (i.e., not a 2019 surprise) — and that Zilis was floating it directly to Musk without alarm; and (iii) the Cerebras passage previews the same Cerebras conflict that surfaces in [[PX 157]] / [[PX 158]] (Sutskever's \"concern is that Tesla has a duty to shareholders…\") and connects to the [[Key Themes#Greg Brockman's contributions (\"zero\")|Brockman March 2017 Cerebras stock purchase]] thread on Day 6. The Dota 5v5 line ties the same conversation to [[PX 90]] (the June 2017 \"10,000 servers\" Microsoft ask). Cross-reference: this email predates [[PX 151|Brockman's August 21, 2017 journal]] by 34 days.\n\n---\n*See also:* [[Day 8|Day 8 digest]] · [[Shivon Zilis]] · [[Greg Brockman]] · [[Ilya Sutskever]] · [[PX 157]] · [[PX 90]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-631", "exhibit": "DX 631", "party": "Defendants", "type": "Email thread", "admitted_trial_day": "Day 3 (April 29, 2026)", "uploaded_box_pt": "2026-04-29T15:37:43", "uploader": "Morrison Foerster", "pages": 2, "size_bytes": 433614, "source_pdf": "DX-631.pdf", "pdf_url": "https://media.mts-in.com/DX-631.pdf", "body_markdown": "# DX 631 — July 21, 2017 Musk-Brockman-Sutskever-Schulman email thread: Brockman's three-stage roadmap \"AI research non-profit (through end of 2017) → AI research + hardware for-profit (starting 2018) → Government project\" + Musk's Founder Series Tesla Model 3 gifts\n\n> The contemporaneous record of Greg Brockman, with Ilya Sutskever and John Schulman copied, telling Musk the path \"must be\" a transition from nonprofit to for-profit by 2018 — and Musk responding affirmatively (\"Cool, that turned out to be an extremely important dinner!\") while gifting Founder Series Tesla Model 3s to the OpenAI co-founders \"in appreciation for what you've done to get OpenAI to where it is today.\"\n\n## Document type\n**Email thread, plain text, six messages.** A reply chain spanning July 21, 2017 1:25 AM PDT through July 21, 2017 6:27 PM PDT. Subject: \"Re: Beijing Wants A.I. to Be Made in China by 2030 - NYTimes.com.\" Participants: Elon Musk <erm@spacex.com>, Greg Brockman <gdb@openai.com>, Ilya Sutskever <ilyasu@openai.com>, John Schulman <joschu@openai.com>, Wojciech Zaremba <woj@openai.com>. Bates 2024MUSK-0006107–6108.\n\n## Logistics\n- **Trial admission:** Day 3 (April 29, 2026) — Musk direct concludes; cross begins. The exhibit is the defense's headline document for the proposition that Musk himself was the one who, in mid-2017, was being told (and tacitly agreeing) that OpenAI's path forward required a for-profit subsidiary — and his response was to gift Tesla Model 3s, not to object. Tied to [[Key Themes]] § \"'Unequivocal control' (the 2017 negotiation breakdown).\"\n- **Box upload:** 2026-04-29 15:37:43 PT — Day 3 mid-afternoon batch (clustered with DX 507, DX 691, DX 869, DX 1025).\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~424 KB, 2 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `631 .pdf`.\n\n## Transcribed text\n\n> **From:** John Schulman <joschu@openai.com>\n> **To:** Elon Musk <erm@spacex.com>\n> **Cc:** Greg Brockman <gdb@openai.com>, Ilya Sutskever <ilyasu@openai.com>, Wojciech Zaremba <woj@openai.com>\n> **Subject:** Re: Beijing Wants A.I. to Be Made in China by 2030 - NYTimes.com\n> **Date:** Fri, 21 Jul 2017 18:27:49 -0700\n>\n> Wow, thanks so much, Elon, and thanks Ilya for recommending me for this honor.\n> I feel like I don't deserve this, but I love driving the Tesla enough that I can't bring myself to decline it.\n>\n> John\n\n> **From:** Elon Musk <erm@spacex.com>\n> **To:** Greg Brockman <gdb@openai.com>\n> **Cc:** Ilya Sutskever <ilyasu@openai.com>, John Schulman <joschu@openai.com>, Wojciech Zaremba <woj@openai.com>\n> **Subject:** Re: Beijing Wants A.I. to Be Made in China by 2030 - NYTimes.com\n> **Date:** Fri, Jul 21, 2017 at 3:38 PM\n>\n> Cool, that turned out to be an extremely important dinner!\n>\n> The road ahead will definitely be vastly more difficult than the past, but extremely exciting.\n\n> **From:** Greg Brockman <gdb@openai.com>\n> **To:** Elon Musk <erm@spacex.com>\n> **Cc:** Ilya Sutskever <ilyasu@openai.com>, John Schulman <joschu@openai.com>, Wojciech Zaremba <woj@openai.com>\n> **Subject:** Re: Beijing Wants A.I. to Be Made in China by 2030 - NYTimes.com\n> **Date:** Fri, Jul 21, 2017 at 2:41 PM\n>\n> Also, just realized today is the **2 year anniversary since our initial dinner at the Rosewood in 2015**. Off to a good start but the hard part is just starting.\n>\n> - gdbr\n\n> **From:** Elon Musk <erm@spacex.com>\n> **To:** Ilya Sutskever <ilyasu@openai.com>\n> **Cc:** Greg Brockman <gdb@openai.com>, John Schulman <joschu@openai.com>, Wojciech Zaremba <woj@openai.com>\n> **Subject:** Re: Beijing Wants A.I. to Be Made in China by 2030 - NYTimes.com\n> **Date:** Fri, Jul 21, 2017 at 2:21 PM\n>\n> You're most welcome!\n\n> **From:** Greg Brockman <gdb@openai.com>\n> **To:** Elon Musk <erm@spacex.com>\n> **Cc:** Ilya Sutskever <ilyasu@openai.com>, John Schulman <joschu@openai.com>, Wojciech Zaremba <woj@openai.com>\n> **Subject:** Re: Beijing Wants A.I. to Be Made in China by 2030 - NYTimes.com\n> **Date:** Fri, Jul 21, 2017 at 1:54 PM\n>\n> As Ilya said, thank you! This really means a lot.\n>\n> - gdb\n\n> **From:** Ilya Sutskever <ilyasu@openai.com>\n> **To:** Elon Musk <erm@spacex.com>\n> **Cc:** Greg Brockman <gdb@openai.com>, John Schulman <joschu@openai.com>, Wojciech Zaremba <woj@openai.com>\n> **Subject:** Re: Beijing Wants A.I. to Be Made in China by 2030 - NYTimes.com\n> **Date:** Fri, Jul 21, 2017 at 1:47 PM\n>\n> Wow, thank you! This is a very kind gesture that we greatly appreciate. Yes, Wojciech definitely deserves it as much as we do, as does +John Schulman.\n\n> **From:** Elon Musk <erm@spacex.com>\n> **To:** Greg Brockman <gdb@openai.com>, Ilya Sutskever <ilyasu@openai.com>\n> **Subject:** Re: Beijing Wants A.I. to Be Made in China by 2030 - NYTimes.com\n> **Date:** Fri, Jul 21, 2017 at 1:25 PM\n>\n> By the way, in appreciation for what you've done to get OpenAI to where it is today, **I would like to give you each a Founder Series Model 3**. These are the earliest cars produced and are not available to the public.\n>\n> If there are a few others at OpenAI who you think also really deserve one (maybe Wojciech?), I am happy to do the same for them.\n>\n> Hope you like it!\n>\n> Elon\n\n> **From:** Greg Brockman <gdb@openai.com>\n> **To:** Elon Musk <erm@spacex.com>\n> **Cc:** Ilya Sutskever <ilyasu@openai.com>\n> **Subject:** Re: Beijing Wants A.I. to Be Made in China by 2030 - NYTimes.com\n> **Date:** Fri, Jul 21, 2017 at 11:32 AM\n>\n> **100% agreed. We think the path must be:**\n>\n> **1. AI research non-profit (through end of 2017)**\n> **2. AI research + hardware for-profit (starting 2018)**\n> **3. Government project (when: ??)**\n>\n> Government project is helpful for financial resources, but crucial for security — don't know how to defend against Putin or North Korea otherwise. When ready, we must proactively form the project around us, rather than being slurped in against our will.\n>\n> - gdb\n\n> **From:** Elon Musk <erm@spacex.com>\n> **To:** Greg Brockman <gdb@openai.com>, Ilya Sutskever <ilyasu@openai.com>\n> **Subject:** Beijing Wants A.I. to Be Made in China by 2030 - NYTimes.com\n> **Date:** Fri, Jul 21, 2017 at 3:34 AM\n>\n> They will do whatever it takes to obtain what we develop. Maybe another reason to change course.\n>\n> https://mobile.nytimes.com/2017/07/20/business/china-artificial-intelligence.html?smprod=nytcore-iphone&smid=nytcore-iphone-share&referer=\n\n## Commentary\n\nDX 631 is a defense layer-cake. The chain begins at 3:34 AM PDT with **Musk** forwarding a New York Times article on China's 2030 AI plan and writing \"Maybe another reason to change course\" — Musk himself opening the door to restructuring. Brockman responds at 4:32 AM with a three-stage path that explicitly contemplates an \"**AI research + hardware for-profit (starting 2018)**\" — i.e., the structural conversion plaintiffs now characterize as a stolen charity. **Defense's read:** Musk does not object to the for-profit step; instead, hours later he announces that he is gifting **Founder Series Tesla Model 3s** to Brockman, Sutskever, and (at Sutskever's suggestion) Zaremba and Schulman \"in appreciation for what you've done to get OpenAI to where it is today\" — concluding the thread with \"Cool, that turned out to be an extremely important dinner!\" The defense pairs DX 631 with DX 507 (the Nov 2015 \"standard C corp with a parallel nonprofit\" email) and the August–September 2017 control-fight emails (PX 156 / PX 157) to argue that the for-profit conversion was Musk's plan as much as anyone else's — what changed in September 2017 was **who controls the for-profit**, not whether one would exist. **Plaintiffs' read:** Brockman's three-stage roadmap was an unsolicited proposal that Musk neither endorsed in writing nor implemented; the gift of Founder Series Model 3s was for the *nonprofit* work, and the actual for-profit conversion in 2019 happened after Musk had already been pushed out (see [[PX 151]], [[PX 154]], [[PX 161]], [[Brockman Journal]]). The **\"Wojciech?\"** reference is to Wojciech Zaremba, OpenAI co-founder. See [[Day 3|Day 3 digest]] for the cross use.\n\n---\n*See also:* [[Day 3|Day 3 digest]] · [[DX 507]] · [[PX 157]] · [[Greg Brockman]] · [[Ilya Sutskever]] · [[Brockman Journal]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-632", "exhibit": "DX 632", "party": "Defense (OpenAI/MS)", "type": "Email thread", "admitted_trial_day": "Day 7 (May 5, 2026)", "uploaded_box_pt": "2026-05-05T14:27:00", "uploader": "Morrison Foerster", "pages": 1, "size_bytes": 386408, "source_pdf": "DX-632.pdf", "pdf_url": "https://media.mts-in.com/DX-632.pdf", "body_markdown": "# DX 632 — July 21–22, 2017 Brockman / Musk \"100% agreed\" for-profit-by-2018 email\n\n> Brockman to Musk, on a NYT \"Beijing Wants A.I. to Be Made in China by 2030\" forward: **\"100% agreed. We think the path must be: 1. AI research non-profit (through end of 2017); 2. AI research + hardware for-profit (starting 2018); 3. Government project (when: ??).\"** The defense's clearest pre-PX 157 contemporaneous proof that Musk and Brockman were jointly planning the for-profit conversion two months before the September 2017 negotiation that plaintiffs say broke the trust.\n\n## Document type\n**Email thread, plain text, four messages threaded.** Top message: Emma Gallagher (Tesla, Musk's assistant) to Brockman cc Sutskever, \"Done!\", Sat 7/22/2017 6:44:03 PM, replying to Brockman's 10:54 PM \"Let's also add Sam Altman to the call!\" Threaded below: (a) Gallagher Fri 7/21 7:10 PM (\"elon is in Australia, but I'll add a call to the calendar\"); (b) Brockman 7/21 2:57 PM forwarding Musk's 7/21 1:18 PM \"Let's talk Sat or Sun. I have a tentative game plan that I'd like to run by you\"; (c) Brockman 7/22 4:32 AM (\"100% agreed. We think the path must be: 1/2/3\"); and (d) at the bottom, Musk's original 7/21 3:34 AM forward of the NYT article: \"They will do whatever it takes to obtain what we develop. Maybe another reason to change course.\" Bates OPENAI_MUSK00008719. Production-stamped CONFIDENTIAL.\n\n## Logistics\n- **Trial admission:** Day 7 (May 5, 2026) — Brockman defense direct + Mr. Molo recross + DX 1252; Wu deposition. Used by the defense on Brockman's direct to back-date the for-profit-conversion plan to *July 2017* — i.e., **Musk-led**, before the [[PX 157]] \"final straw\" rupture.\n- **Box upload:** 2026-05-05 14:27:00 PT — Day 7 mid-afternoon batch.\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~377 KB, 1 page.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `0632.pdf`.\n\n## Transcribed text\n\n> **From:** Emma Gallagher <emma@tesla.com>\n> **To:** Greg Brockman <gdb@openai.com>\n> **Cc:** Ilya Sutskever <ilyasu@openai.com>\n> **Subject:** Re: Beijing Wants A.I. to Be Made in China by 2030 - NYTimes.com\n> **Date:** 7/22/2017 6:44:03 PM\n>\n> Done!\n\n> **From:** Greg Brockman <gdb@openai.com>\n> **To:** Emma Gallagher <emma@tesla.com>\n> **Cc:** Ilya Sutskever <ilyasu@openai.com>\n> **Subject:** Re: Beijing Wants A.I. to Be Made in China by 2030 - NYTimes.com\n> **Date:** Fri, Jul 21, 2017 at 10:54 PM\n>\n> Let's also add Sam Altman to the call!\n>\n> - gdb\n\n> **From:** Emma Gallagher <emma@tesla.com>\n> **To:** Greg Brockman <gdb@openai.com>\n> **Cc:** Ilya Sutskever <ilyasu@openai.com>\n> **Subject:** Re: Beijing Wants A.I. to Be Made in China by 2030 - NYTimes.com\n> **Date:** Fri, Jul 21, 2017 at 7:10 PM\n>\n> Hi Guys – elon is in Australia, but I'll add a call to the calendar as soon as I have a time. Will likely be tomorrow night our time.\n>\n> Thanks,\n> Emma\n\n> **From:** Greg Brockman <gdb@openai.com>\n> **To:** Emma Gallagher <emma@tesla.com>\n> **Cc:** Ilya Sutskever <ilyasu@openai.com>\n> **Subject:** Re: Beijing Wants A.I. to Be Made in China by 2030 - NYTimes.com\n> **Date:** 7/21/17, 2:57 PM\n>\n> Please let us know when works best for the call!\n>\n> - gdb\n>\n> ---------- Forwarded message ----------\n\n> **From:** Elon Musk <erm@spacex.com>\n> **To:** Greg Brockman <gdb@openai.com>\n> **Cc:** Ilya Sutskever <ilyasu@openai.com>, Sam Teller <steller@spacex.com>\n> **Subject:** Re: Beijing Wants A.I. to Be Made in China by 2030 - NYTimes.com\n> **Date:** Fri, Jul 21, 2017 at 1:18 PM\n>\n> Let's talk Sat or Sun. I have a tentative game plan that I'd like to run by you.\n\n> **From:** Greg Brockman <gdb@openai.com>\n> **To:** Elon Musk <erm@spacex.com>\n> **Cc:** Ilya Sutskever <ilyasu@openai.com>\n> **Subject:** Re: Beijing Wants A.I. to Be Made in China by 2030 - NYTimes.com\n> **Date:** Sat, Jul 22, 2017 at 4:32 AM\n>\n> 100% agreed. We think the path must be:\n>\n> 1. AI research non-profit (through end of 2017)\n> 2. AI research + hardware for-profit (starting 2018)\n> 3. Government project (when: ??)\n>\n> Government project is helpful for financial resources, but crucial for security – don't know how to defend against Putin or North Korea otherwise. When ready, we must proactively form the project around us, rather than being slurped in against our will.\n>\n> - gdb\n\n> **From:** Elon Musk <erm@spacex.com>\n> **To:** Greg Brockman <gdb@openai.com>, Ilya Sutskever <ilyasu@openai.com>\n> **Subject:** Beijing Wants A.I. to Be Made in China by 2030 - NYTimes.com\n> **Date:** Fri, Jul 21, 2017 at 3:34 AM\n>\n> They will do whatever it takes to obtain what we develop. Maybe another reason to change course.\n>\n> https://mobile.nytimes.com/2017/07/20/business/china-artificial-intelligence.html?smprod=nytcore-iphone&smid=nytcore-iphone-share&referer=\n\n## Commentary\n\nDX 632 is the defense's \"Musk wanted the for-profit too\" smoking gun for the *summer* of 2017. The chronology is the point: on July 21 Musk forwards the NYT China-AI piece and writes \"**Maybe another reason to change course**\"; Brockman responds the next morning \"**100% agreed**\" with a three-step plan whose step 2 — \"**AI research + hardware for-profit (starting 2018)**\" — is exactly the structure plaintiffs would later say was \"stolen.\" When that plan blew up in September 2017, the dispute (per [[PX 157]]) wasn't *whether* to convert but *who would control* the converted entity — see [[Key Themes#\"Unequivocal control\"]]. Brockman on direct used DX 632 to argue the for-profit plan started as a *Musk* idea (\"change course,\" then \"tentative game plan I'd like to run by you\"); on recross, plaintiffs' [[Mr. Molo|Molo]] would have pushed the contrary read that Brockman was over-eagerly drafting Musk's marbles into a for-profit Musk would still ultimately walk away from. The \"Government project... defend against Putin or North Korea\" line is also a useful color thread for the [[Key Themes#AGI safety as the original purpose|safety]] motif, though the court has bracketed AGI-safety testimony.\n\n---\n*See also:* [[Day 7|Day 7 digest]] · [[Greg Brockman]] · [[PX 157]] · [[DX-559]] · [[DX-758]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-638", "exhibit": "DX 638", "party": "Defense", "type": "Contract", "admitted_trial_day": "Day 4 (April 30, 2026)", "uploaded_box_pt": "2026-04-30T15:13:13", "uploader": "Morrison Foerster", "pages": 6, "size_bytes": 577698, "source_pdf": "DX-638.pdf", "pdf_url": "https://media.mts-in.com/DX-638.pdf", "body_markdown": "# DX 638 — Aug 7, 2017 Musk Foundation → Fidelity Charitable Contribution Form (TSLA, 6,000 shares ≈ $2.1M)\n\n> The Aug 7, 2017 Fidelity Charitable contribution form by which **Musk Foundation** transferred **6,000 shares of Tesla Inc. (≈ $2,100,000)** through Morgan Stanley into a Fidelity Charitable Giving Account — one of the donor-advised-fund mechanics behind Musk's 2017 contributions to OpenAI.\n\n## Document type\n**Contract / form** — Fidelity Charitable \"Contribution Form and Letter of Instruction\" (Form 1.851030.117), 6 pages, signed by Musk Foundation, with delivery instructions for transferring TSLA shares from a Morgan Stanley account (broker: Jon Neuhaus, Los Angeles, CA 90071) to Fidelity Charitable Giving Account #...6490. Bates FIDCHAR-OPENAI-000876–000881. Marked HIGHLY CONFIDENTIAL.\n\n## Logistics\n- **Trial admission:** Day 4 (April 30, 2026) — Birchall direct + cross. Used in the [[Key Themes]] §\"donor-advised funds / tax-deduction wrinkle\" line of attack on plaintiffs' \"stole a charity\" framing.\n- **Box upload:** 2026-04-30 15:13:13 PT — Day 4 mid-afternoon Defense batch (uploaded 18 seconds before [[DX 600]]).\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~564 KB, 6 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `638.pdf`.\n\n## Transcribed text\n\n> **FIDELITY Charitable — Contribution Form and Letter of Instruction**\n>\n> **1. Account Holder or Third-Party Contributor**\n> Type: ☒ Third-Party Contributor\n> Name: **Musk Foundation**, Hawthorne, CA 90250, USA\n> Tax Identification Number: ...7507\n>\n> **2. Assets from a Fidelity Investments Account** — *(blank)*\n>\n> **3. Via Check or Wire** — *(blank)*\n>\n> **5. Investment Pool Selection** — ☒ Allocate this contribution in the same percentages as my current Giving Account balance.\n>\n> **LETTER OF INSTRUCTION**\n>\n> **6. Contribute Securities or Mutual Funds Held at a Financial Institution Other than Fidelity**\n> Method: ☒ I will initiate the transfer of assets by completing the following three steps: 1) Contact financial institution directly to obtain required procedures and forms, and provide it with the delivery instructions in Section 7. Deliver to DTC Clearing 226. 2) Complete a separate form for each transfer from separate financial institutions. 3) Complete and fax this form to Fidelity Charitable: 877-665-4274.\n>\n> **TO:** Fidelity Charitable Giving Account Number: **6490**\n>\n> **FROM:** Delivering Firm Name: **Morgan Stanley** — Broker Name: **Jon Neuhaus** — Date: **04/07/2017** — City/State/ZIP: **Los Angeles, CA 90071** — Account Number at Delivering Firm: ...5048\n>\n> **TRANSFER:**\n> Number of Shares: **6,000** | Approx. Dollar Amount (if known): **$2,100,000** | OF: Description of Assets/Name of Security: **TESLA INC** | Symbol: **TSLA**\n>\n> **7. Signatures and Dates**\n> Account Holder Name or Third-Party Contributor Name: **Musk Foundation**\n> Date: **8/7/17**\n> Signature: [signed]\n>\n> **Delivery Options**\n> Mail this form to Fidelity Charitable: Regular mail — Fidelity Charitable, P.O. Box 770001, Cincinnati, OH 45277-0053. Overnight mail — Fidelity Charitable, 100 Crosby Parkway KC1D-FC5, Covington, KY 41015-9325. OR Fax the form: 877-665-4274.\n>\n> Delivery instructions for contributing assets held outside Fidelity Investments — DTC-Eligible Securities: Deliver to DTC Clearing 226, Account Number: Z97-000442, FBO: Fidelity Charitable Giving Account Number: [redacted]. Wire instructions: Wire to: JPMorgan Chase Bank, New York, NY; ABA Number: 021000021; For credit to: National Financial Services LLC, Account: 066196-221; FBO: Fidelity Charitable, Z97-000442; Fidelity Charitable Giving Account Number: __. Please note: Fidelity Charitable will not initiate wire transfers.\n\n> [Page 6 — Fidelity Charitable disclosures: \"Fidelity Charitable does not provide legal or tax advice. … Charitable contributions of capital gain property held for more than one year are usually deductible at fair market value. Deductions for capital gain property held for one year or less are usually limited to cost basis. Consult an attorney or tax advisor regarding your specific legal or tax situation. … Fidelity Charitable is the brand name for Fidelity® Charitable Gift Fund, an independent public charity with a donor-advised fund program.\"]\n\n## Commentary\n\nDX 638 is defense's clean documentary basis for the **donor-advised-fund / tax-deduction** point on Birchall's cross (see [[Key Themes]] §\"donor-advised funds / tax-deduction wrinkle\"): every Musk-to-OpenAI dollar passed through a 501(c)(3) intermediary (Vanguard Charitable in [[PX 60]] / Fidelity Charitable here), and once Musk transferred to a DAF, \"**he could not take the money back — he could only direct it.**\" The 6,000-share TSLA contribution at ~$350/share (Aug 2017) is roughly the amount of one quarterly OpenAI grant cycle for the Musk Foundation account. The \"Musk Foundation\" entity name and Hawthorne, CA address (the SpaceX HQ) tie back to Birchall's role as Managing Director of the Musk Family Office in [[PX 79]]. Plaintiffs' standing answer is that even DAF contributions remain subject to charitable-trust principles under California law; defense's answer is that Musk had no surviving personal claim to the dollars. Pair with [[PX 60]] for the Vanguard Charitable side of the same flow architecture.\n\n---\n*See also:* [[Day 4|Day 4 digest]] · [[PX 60]] · [[PX 79]] · [[Jared Birchall]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-642", "exhibit": "DX 642", "party": "Defense", "type": "Email thread", "admitted_trial_day": "Day 3 (April 29, 2026)", "uploaded_box_pt": "2026-04-29T15:37:43", "uploader": "Morrison Foerster", "pages": 1, "size_bytes": 379989, "source_pdf": "DX-642.pdf", "pdf_url": "https://media.mts-in.com/DX-642.pdf", "body_markdown": "# DX 642 — Aug 11–12, 2017 Musk → founders, \"This is the triggering event\" (post-Dota / Haunted Mansion meeting)\n\n> A one-page reply-chain email from August 11–12, 2017 — Musk's \"Time to make the next step for OpenAI. **This is the triggering event.**\" message after OpenAI's Dota 2 1v1 victory, scheduling a meet at Musk's \"haunted mansion\" near SF the next afternoon with Brockman, Sutskever, Altman, Teller, and Zilis.\n\n## Document type\n**Email thread, plain text, multi-message reply chain.** Brockman's \"Thanks for the invite\" reply on top of a fully quoted scheduling chain among Musk, Altman, Brockman, Sutskever, Zilis, and Teller. From: Greg Brockman <gdb@openai.com>; subject \"Re: Tomorrow afternoon\"; date Sat, 12 Aug 2017 07:37:47 -0700. Bates 2024MUSK-0005629.\n\n## Logistics\n- **Trial admission:** Day 3 (April 29, 2026 — Musk direct concludes; cross begins). Per the Day 3 transcript table: \"DX 642 | Aug. 11–12, 2017 email 'Time to make the next step for OpenAI. This is the triggering event'; Haunted Mansion meeting\" `(042926TT.txt:3473–3552)`. Plaintiffs cross used it: \"Q. 'And the triggering event was the Dota victory; correct?' / A. 'Probably.'\" `(042926TT.txt:3528–3535)`.\n- **Box upload:** 2026-04-29 15:37:43 PT — Day 3 late-afternoon defense batch.\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~371 KB, 1 page.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `642 .pdf` (note trailing space in filename).\n\n## Transcribed text\n\n> **From:** Greg Brockman <gdb@openai.com>\n> **To:** Elon Musk <erm@spacex.com>\n> **Cc:** Shivon Zilis <shivon@openai.com>, Ilya Sutskever <ilyasu@openai.com>, Sam Altman <shga@ycombinator.com>, Sam Teller <steller@spacex.com>\n> **Subject:** Re: Tomorrow afternoon\n> **Date:** Sat, 12 Aug 2017 07:37:47 -0700\n>\n> Thanks for the invite — would have come if around!\n>\n> - gdb\n\n> **From:** Elon Musk <erm@spacex.com>\n> **To:** Shivon Zilis <shivon@openai.com>\n> **Cc:** Ilya Sutskever <ilyasu@openai.com>, Sam Altman <shga@ycombinator.com>, Greg Brockman <gdb@openai.com>, Sam Teller <steller@spacex.com>\n> **Subject:** Re: Tomorrow afternoon\n> **Date:** Fri, Aug 11, 2017 at 10:15 PM\n>\n> Sounds good, let's aim for 3pm. We can probably meet at the haunted mansion I just bought near SF. It's kinda crazy and weird and will have party carnage, but it might make for a good backdrop.\n>\n> Btw, anyone in town tonight is welcome to come to the party. House is about 15 mins south of SF.\n\n> **From:** Shivon Zilis <shivon@openai.com>\n> **To:** Elon Musk <erm@spacex.com>\n> **Cc:** Ilya Sutskever <ilyasu@openai.com>, Sam Altman <shga@ycombinator.com>, Greg Brockman <gdb@openai.com>, Sam Teller <steller@spacex.com>\n> **Subject:** Re: Tomorrow afternoon\n> **Date:** Fri, Aug 11, 2017 at 10:02 PM\n>\n> Everyone says they are available in Bay Area the window of 3-6 if anything in there works.\n\n> **From:** Ilya Sutskever <ilyasu@openai.com>\n> **To:** Elon Musk <erm@spacex.com>\n> **Cc:** Shivon Zilis <shivon@openai.com>, Sam Altman <shga@ycombinator.com>, Greg Brockman <gdb@openai.com>, Sam Teller <steller@spacex.com>\n> **Subject:** Re: Tomorrow afternoon\n> **Date:** Fri, Aug 11, 2017 at 9:52 PM\n>\n> Can do, but would need to leave at 7pm.\n\n> **From:** Greg Brockman <gdb@openai.com>\n> **To:** Elon Musk <erm@spacex.com>\n> **Cc:** Shivon Zilis <shivon@openai.com>, Ilya Sutskever <ilyasu@openai.com>, Sam Altman <shga@ycombinator.com>, Sam Teller <steller@spacex.com>\n> **Subject:** Re: Tomorrow afternoon\n> **Date:** Fri, Aug 11, 2017 at 9:38 PM\n>\n> Yep -- moving my flight up from Sunday to tomorrow.\n\n> **From:** Sam Altman <shga@ycombinator.com>\n> **To:** Elon Musk <erm@spacex.com>\n> **Cc:** Shivon Zilis <shivon@openai.com>, Ilya Sutskever <ilyasu@openai.com>, Greg Brockman <gdb@openai.com>, Sam Teller <steller@spacex.com>\n> **Subject:** Re: Tomorrow afternoon\n> **Date:** Fri, Aug 11, 2017 at 9:33 PM\n>\n> yes i can meet\n\n> **From:** Elon Musk <erm@spacex.com>\n> **To:** Sam Altman <shga@ycombinator.com>, Greg Brockman <gdb@openai.com>, Ilya Sutskever <ilyasu@openai.com>\n> **Cc:** Shivon Zilis <shivon@openai.com>, Sam Teller <steller@spacex.com>\n> **Subject:** Tomorrow afternoon\n> **Date:** Fri, Aug 11, 2017 at 9:18 PM\n>\n> I'm in SF this weekend for a friend's wedding. Are you guys able to meet or do a conf call tomorrow afternoon?\n>\n> Time to make the next step for OpenAI. **This is the triggering event.**\n>\n> --\n> Sent from mobile\n\n## Commentary\n\nDX 642 is the **August 2017 inflection-point email** — Musk's own \"**This is the triggering event**\" language one day after OpenAI won the Dota 2 1v1 championship at The International. Defense uses it for the [[Key Themes|\"three phases\"]] timeline rebuttal: Musk himself, in his own words, treated August 2017 as the triggering moment for moving OpenAI to a **for-profit** structure, undermining the plaintiffs' framing that the for-profit conversion came as a betrayal. Plaintiffs' cross got Musk to concede on the stand that \"the triggering event was the Dota victory; correct? — Probably.\" `(042926TT.txt:3528)`. The \"haunted mansion … party carnage … good backdrop\" line is a small Musk-voice authenticity tell that makes the August 12, 2017 meeting concrete — and the meeting itself is the prelude to the cap-table-and-control negotiations that produced [[PX 153|PX 153 (Sept 11 cap table)]] and the [[Brockman Journal|Brockman journal entries]] of August 18 ([[PX-1250-16]] / [[PX 151|PX 154]]) and August 21 (PX 151). See [[Day 3|Day 3 digest]] for the cross use.\n\n---\n*See also:* [[Day 3|Day 3 digest]] · [[Elon Musk]] · [[Greg Brockman]] · [[Brockman Journal]] · [[PX 151]] · [[PX 153]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-643", "exhibit": "DX 643", "party": "Defense", "type": "Email", "admitted_trial_day": "Day 8 (May 6, 2026)", "uploaded_box_pt": "2026-05-06T14:57:05", "uploader": "Morrison Foerster", "pages": 1, "size_bytes": 386343, "source_pdf": "DX-643.pdf", "pdf_url": "https://media.mts-in.com/DX-643.pdf", "body_markdown": "# DX 643 — Aug 12, 2017 Zilis \"Decisions and Notes\" email — OpenAI for-profit \"switch in next couple of weeks\"\n\n> Shivon Zilis's August 12, 2017 9:01 PM internal report to Sam Teller and Emma Gallagher summarizing a 5:30–7:30 meeting with Musk: **\"OpenAI: switch to for profit in next couple of weeks (woah, fast!)\"** — used by defense to anchor Musk's own contemporaneous knowledge of and participation in the for-profit pivot, weeks before the [[PX 157]] \"final straw\" exchange.\n\n## Document type\n**Internal email, single page, plain-text bullet notes.** From Shivon Zilis <shivon@tesla.com> to Sam Teller <steller@tesla.com> and Emma Gallagher <emma@tesla.com>, sent Sat, 12 Aug 2017 21:01:35, subject \"Decisions and Notes.\" Production-stamped HIGHLY CONFIDENTIAL — AEO; Bates TESLA_000002038. Document ID DX-0643.\n\n## Logistics\n- **Trial admission:** Day 8 (May 6, 2026) — Murati live; Zilis live; Toner video begins. Used during the Zilis live examination as part of the defense's \"Musk knew, Musk participated\" sequence (see [[Key Themes#\"Bury this in Tesla for stealth advantage\"]] and the Aug 2017 \"ironclad agreement\" / funding-freeze sequence at `5/6/2026 Testimony @ ~11:14 PT`).\n- **Box upload:** 2026-05-06 14:57:05 PT — Day 8 mid-afternoon batch (clustered with DX 707, DX 728, DX 758, DX 827, DX 925).\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~377 KB, 1 page.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `0643.pdf`.\n\n## Transcribed text\n\n> **From:** Shivon Zilis <shivon@tesla.com>\n> **To:** Sam Teller <steller@tesla.com>, Emma Gallagher <emma@tesla.com>\n> **Subject:** Decisions and Notes\n> **Date:** 8/12/2017 9:01:35 PM\n>\n> Hey guys!\n>\n> Productive meeting. Ran 5:30-7:30ish I think(?) so contentful discussion.\n>\n> **OpenAI:**\n> - switch to for profit in next couple of weeks (woah, fast!)\n> - retain .org, add .com\n> - keep a subset of people at OpenAI non-profit, likely safety and regulation oriented, but E wanted some coders there too\n> - unclear where DoTA 5v5 IP will reside yet\n> - messaging around for profit is need equity to retain true AI talent, need money for hardware\n> - all hands on deck 5v5\n> - **Elon and Sam A to lock down server agreement with Satya, unclear who will take lead but likely Sam** Context is after E and Satya talked, Satya's team offered a shitty deal. This may involve both Jensen and Satya to get to the right deal but likely Satya first.\n> - Thinks OpenAI team way too enamored with Cerebras. Isn't opposed to something in time but thinks they need to be more measured and less dupeable.\n> - Ran my Andrew Feldman meeting by him and he was cool with it so will report back on that Monday.\n>\n> **Neuralink:**\n> - Hm. Sam A thinks Max is a disaster (said those words) and has duped me and everyone else (said in front of Elon). This clearly has transcriptic roots but he told me if I don't think Max is gunning to be CEO he's duped me. (Please don't repeat outside of Elon)\n> - **Emma — Elon confirmed he 100% wants to do movie. I tried to offer option out but his take was he wants to see the new cut anyway and is concerned that he doesn't have enough feedback yet**\n> - We talked org structure. He knows it's a problem and will think on it. No short term to do.\n> - Wants to figure out how to productively spend more time since he thinks it's only way to AGI future (this is tricky because both he doesn't have the time and Neuralink still doesn't know how to use him well)\n>\n> **Neil Strauss:**\n> - **He says no to Neuralink, yes to OpenAI. Who should be setting expectations with Nick?**\n\n## Commentary\n\nDX 643 is one of the cleanest pieces of evidence the defense has for Musk's **personal knowledge and assent** to the OpenAI for-profit pivot in real time — written by the very Tesla-side staffer (Zilis) plaintiffs in the underlying case relied upon as Musk's eyes-and-ears at OpenAI. Three lines in particular weaponize: **\"switch to for profit in next couple of weeks (woah, fast!)\"**, **\"messaging around for profit is need equity to retain true AI talent, need money for hardware\"**, and **\"Elon and Sam A to lock down server agreement with Satya\"** — the last of which seeds the August 2017 Microsoft/Azure compute negotiation that ultimately becomes the structural backbone of the [[Key Themes#\"Captured by Microsoft\"|Microsoft-capture]] story. The exhibit reads as the *baseline* against which Brockman's [[PX 151]] \"**This is the only chance we have to get out from under Elon**\" journal entry of August 21, 2017 lands — Musk on August 12 was actively planning the same conversion the founders, nine days later, would resolve to do without him. It is the contemporaneous companion to the Zilis \"ironclad agreement to not have Elon or anyone have absolute control of AGI\" note ([[Key Themes]]) and the August 20 funding-freeze email — and runs in opposite directions on plaintiffs' \"stole a charity\" framing.\n\n---\n*See also:* [[Day 8|Day 8 digest]] · [[Shivon Zilis]] · [[PX 151]] · [[PX 157]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-644", "exhibit": "DX 644", "party": "Defense", "type": "Text messages", "admitted_trial_day": "Day 8 (May 6, 2026)", "uploaded_box_pt": "2026-05-06T14:57:06", "uploader": "Morrison Foerster", "pages": 4, "size_bytes": 655164, "source_pdf": "DX-644.pdf", "pdf_url": "https://media.mts-in.com/DX-644.pdf", "body_markdown": "# DX-644 — Aug 15, 2017 Zilis/Teller/Gallagher texts: structuring Musk's OpenAI visit\n\n> Three-way logistics texting between Shivon Zilis, Sam Teller, and Emma Gallagher about scheduling Musk's August 15, 2017 OpenAI visit — including a candid line that the meeting wasn't just Cerebras but **\"the full discussion on immediate next step on conversion to for profit.\"**\n\n## Document type\n**Text messages** (Apple Short Message Report extract). One conversation, 48 messages on 8/15/2017, three participants: Sam Teller (Musk's chief of staff), Shivon Zilis, Emma Gallagher (Musk-side scheduler). Bates ZILIS-0002719–0002722.\n\n## Logistics\n- **Trial admission:** Day 8 (May 6, 2026) — Murati live; Zilis live; Toner video begins.\n- **Box upload:** 2026-05-06 14:57:06 PT — Day-8 batch, clustered with DX 669, DX 708, DX 926, DX 1145.\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~655 KB, 4 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `0644.pdf`.\n\n## Transcribed text\n\n> **Short Message Report**\n> Conversations: 1 · Total Messages: 48 · Date Range: 8/15/2017\n> Participants: Sam Teller, Emma Gallagher, Shivon Zilis\n>\n> **Emma Gallagher** [3:47 PM]: Hey sam will you grab shivon when you arrive\n> **Emma Gallagher** [3:47 PM]: They are parked on a remote ramp\n> **Sam Teller** [3:47 PM]: Ok\n> **Sam Teller** [3:48 PM]: I don't want to valet the Model 3 tho\n> **Emma Gallagher** [3:48 PM]: They will pull the plane up when he is coming\n> **Sam Teller** [3:49 PM]: To normal spot?\n> **Sam Teller** [3:50 PM]: I'm gonna park at main building and they can cart us if needed\n> **Emma Gallagher** [3:54 PM]: Ok\n> **Shivon Zilis** [4:03 PM]: Oh it's close'\n> **Emma Gallagher** [6:06 PM]: Try and ask if Neil should still come if you can\n> **Emma Gallagher** [6:07 PM]: Would be good to cancel his flight sooner rather than later if needed\n> **Sam Teller** [6:32 PM]: Ok for Neil to come\n> **Shivon Zilis** [6:37 PM]: He is saying as of now he will sit in for movie even if just checks email\n> **Emma Gallagher** [6:39 PM]: ok!\n> **Sam Teller** [6:53 PM]: Emma have you told Neil about the movie yet? Good for him to be there for it\n> **Emma Gallagher** [6:53 PM]: No I didn't share specifics yet on what he'd be participating in\n> **Sam Teller** [6:54 PM]: K I'm texting with him. What is his plan?\n> **Sam Teller** [6:54 PM]: (Where and when has been told to go?)\n> **Emma Gallagher** [6:57 PM]: I have been in touch with him assistant who planned for him to fly in and be in SF around 7pm\n> **Emma Gallagher** [6:57 PM]: Told her that Elon would be there around 7:30pm likely and that he was welcome to arrive early.\n> **Sam Teller** [6:57 PM]: Ok cool\n> **Emma Gallagher** [6:57 PM]: Meetings he'll join:\n> **Emma Gallagher** [6:57 PM]: 7:15 PM: OpenAI (DoTA) w/ Neil\n> **Emma Gallagher** [6:57 PM]: 8:30 PM: Movie at Neuralink\n> **Emma Gallagher** [6:58 PM]: I could also swap and do neuralink first so there is less of a gap for Neil\n> **Sam Teller** [6:58 PM]: Ok - Shivon will nail down the Neuralink hour and see what we can include him in if anything. Otherwise can do AI safety chat with Dario and Jack.\n> **Emma Gallagher / Sam Teller** [6:59 PM]: Ok great\n> **Shivon Zilis** [6:59 PM]: Do we think we can legitimately do off the record? It's going to be about monkeys and sewing machines\n> **Shivon Zilis** [8:44 PM]: We should find a way to make Greg / Sam A meeting 30mins. Do you think best way to do that to make Neuralink 45mins?\n> **Sam Teller** [8:46 PM]: Ok\n> **Emma Gallagher** [8:47 PM]: Does the cerebral or Dota section need the extra time?\n> **Emma Gallagher** [8:48 PM]: Also, I think I can just get us out of here a little sooner\n> **Sam Teller** [8:52 PM]: Cerebras shouldn't need much time right S?\n> **Shivon Zilis** [8:53 PM]: It's not just Cerebras… **it's the full discussion on immediate next step on conversion to for profit**\n> **Shivon Zilis** [8:53 PM]: Elon just said to invite Sam A tonight to discuss\n> **Shivon Zilis** [8:54 PM]: DoTA is good to stay at 15 mins\n> **Sam Teller** [8:54 PM]: I know. Saying there isn't really much C stuff to cover\n> **Shivon Zilis** [8:55 PM]: Yeah we can relabel that OpenAI discussion if we want\n> **Emma Gallagher** [8:56 PM]: ok l\n> **Emma Gallagher** [8:56 PM]: i'll make it 30 and invite sam Altman\n> **Shivon Zilis** [8:56 PM]: Thank you!\n> **Emma Gallagher** [8:58 PM]: We just have greg and Altman on there for now\n> **Emma Gallagher** [8:58 PM]: Who else?\n> **Shivon Zilis** [8:58 PM]: Ilya may dial in\n> **Shivon Zilis** [8:58 PM]: Is in Toronto\n> **Emma Gallagher** [8:58 PM]: Ok I'll add\n> **Shivon Zilis** [9:23 PM]: Can we go ahead with Neil / Dario on AI safety while Elon is in Neuralink\n> **Sam Teller** [9:28 PM]: Ok\n\n## Commentary\n\nDX 644 is small but evidentiary gold for the defense's \"Musk drove the for-profit conversion\" theme on [[Day 8|Day 8]]. The line that does the work is Zilis's correction at 8:53 PM: the August 15, 2017 OpenAI agenda item wasn't a Cerebras update — it was **\"the full discussion on immediate next step on conversion to for profit\"** — and Musk himself directed inviting Sam Altman that night to discuss it. This is contemporaneous, on Musk's side of the house (Zilis worked for Musk at Neuralink/Tesla), and predates the [[PX 157]] \"final straw\" by five weeks. The exhibit also drops casual references to \"movie at Neuralink\" and \"monkeys and sewing machines\" — a rare glimpse of the Musk operating cadence around the time the [[Brockman Journal]] entries in [[PX 151]] were being written.\n\n---\n\n*See also:* [[Day 8|Day 8 digest]] · [[Shivon Zilis]] · [[PX 157]] · [[PX 151]] · [[Key Themes]] · [[Brockman Journal]]\n"}
{"exhibit_id": "DX-646", "exhibit": "DX 646", "party": "Defense", "type": "Email", "admitted_trial_day": "Day 4 (April 30, 2026)", "uploaded_box_pt": "2026-04-30T15:13:13", "uploader": "Morrison Foerster", "pages": 1, "size_bytes": 397667, "source_pdf": "DX-646.pdf", "pdf_url": "https://media.mts-in.com/DX-646.pdf", "body_markdown": "# DX 646 — Aug 16, 2017 Zilis → Birchall, \"OpenAI conversion\" notes\n\n> Shivon Zilis's status email recapping a Musk-side meeting on the proposed OpenAI for-profit \"**conversion**\" — including \"**Elon notes from this morning**\": file C-Corp immediately, market-cap structure proportional to donations, \"**Elon wants to have control to prevent this from going squirrelly**,\" and a $100M-on-$1B initial round target.\n\n## Document type\n**Email thread, plain text, three messages.** Zilis's August 16, 2017 7:26 AM \"Thank you, Jared!\" reply on top of Birchall's August 16, 2017 12:16 AM YTD-contributions update, atop Zilis's August 15, 2017 4:22 PM original \"OpenAI conversion\" memo. Bates OPENAI_MUSK00003808. Document ID DX-0646.\n\n## Logistics\n- **Trial admission:** Day 4 (April 30, 2026) — used by defense during Birchall direct/cross to anchor the August 2017 control debate immediately preceding the September 20 \"final straw\" email ([[PX 157]]). No prior wiki reference; admission day inferred from Box upload date.\n- **Box upload:** 2026-04-30 15:13:13 PT — Day 4 mid-afternoon batch (clustered with DX 539, DX 1156, DX 1285).\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~388 KB, 1 page.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `646.pdf`.\n\n## Transcribed text\n\n> **From:** Shivon Zilis <shivon@openai.com>\n> **To:** Jared Birchall <jbirchall@muskfoundation.org>\n> **Cc:** Sam Teller <steller@openai.com>\n> **Subject:** Re: OpenAI conversion\n> **Date:** 8/16/2017 7:26:37 AM\n>\n> Thank you, Jared!\n\n> **From:** Jared Birchall <jbirchall@muskfoundation.org>\n> **To:** Shivon Zilis <shivon@openai.com>\n> **Cc:** Sam Teller <steller@openai.com>\n> **Subject:** Re: OpenAI conversion\n> **Date:** Aug 16, 2017, at 12:16 AM\n>\n> Still waiting for Chris to confirm Sam's YTD contributions, but in 2016 E gave $15.4M and Sam gave $3.7M. E has given ~$16M this year and is on track to give over $22M. I'll have all the numbers by tomorrow morning.\n\n> **From:** Shivon Zilis <shivon@openai.com>\n> **To:** Jared Birchall <jbirchall@muskfoundation.org>\n> **Cc:** Sam Teller <steller@openai.com>\n> **Subject:** OpenAI conversion\n> **Date:** Tue, Aug 15, 2017 at 4:22 PM\n>\n> Jared,\n>\n> Great to chat with you earlier on all of the OpenAIs.\n>\n> **Since we chatted:**\n> - Confirmed with Greg that he and Sam Altman have not executed on anything yet.\n> - Found out Sam Altman's mental timeline here in perhaps a month, Elon's seems to be fairly soon (next couple of weeks?)\n> - I asked Greg to make sure that he, Chris, and Sam A keep the three of us and, when appropriate, Elon in the loop as new information emerges.\n>\n> **Elon notes from this morning:**\n> - File for C Corp immediately (in process - Jared - he would like OpenAI Corporation)\n> - Jared to figure out invested amount and financials (in process - Jared)\n> - Wants to devise a market cap structure proportional to how much each entity has donated, including people at OpenAI (need to confirm that this is ok from a legal perspective? can non-profit donations have bearing on a for profit capitalization?)\n> - **Elon wants to have control to prevent this from going squirrelly**\n> - Rough idea for initial fundraising structure would be 100M on 1B with some top VCs\n> - We will retain both .com and .org for OpenAI\n> - .org to be more about safety and policy, .com core research and applications\n>\n> **Unknowns:**\n> - Leadership of new entity (Greg 100% doesn't want to run it)\n> - Cerebras? (we will likely learn a bit more during Greg / Elon meeting tonight)\n\n## Commentary\n\nDX 646 is a single-page bombshell for the defense. Five weeks before [[PX 157]]'s \"**final straw**\" exchange, Zilis is recording in writing — to the Musk family office's managing director — that **Musk himself** wanted to: (i) \"**File for C Corp immediately**\" with the corporate name **\"OpenAI Corporation\"**; (ii) build a market-cap structure based on contributions (\"**including people at OpenAI**\" — i.e., founder equity); (iii) \"**have control to prevent this from going squirrelly**\"; and (iv) raise **$100M on a $1B post**. Defense uses this email to argue that the conversion to a for-profit was not Altman's machination foisted on a charitable Musk — it was Musk's idea, laid out in a numbered list by Musk's own conduit at OpenAI, on Musk's preferred timeline (\"**Elon's seems to be fairly soon (next couple of weeks?)**\"). The donation totals in Birchall's middle paragraph ($15.4M / $3.7M for 2016; $16M YTD for 2017) also line up with later defense exhibits to rebut plaintiffs' \"Musk gave $100M+\" framing. Cross-reference: Sutskever and Brockman's [[PX 157]] line (\"**you've shown to us that absolute control is extremely important to you**\") tracks the \"Elon wants to have control\" item here almost word for word; and the Cerebras line ties to [[DX 626]].\n\n---\n*See also:* [[Day 4|Day 4 digest]] · [[Shivon Zilis]] · [[Jared Birchall]] · [[PX 157]] · [[DX 626]] · [[Phases One Two Three]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-651", "exhibit": "DX 651", "party": "Defense", "type": "Email", "admitted_trial_day": "Day 8 (May 6, 2026)", "uploaded_box_pt": "2026-05-06T14:57:06", "uploader": "Morrison Foerster", "pages": 2, "size_bytes": 444716, "source_pdf": "DX-651.pdf", "pdf_url": "https://media.mts-in.com/DX-651.pdf", "body_markdown": "# DX 651 — Aug 20–21, 2017 Zilis → Birchall \"Fwd: OpenAI possible notes\" (Greg/Ilya/Sam negotiation read)\n\n> Shivon Zilis's August 21, 2017 forward to Jared Birchall of her own August 20, 2017 11:40 PM strategy memo to Sam Teller — a candid read of where Greg Brockman, Ilya Sutskever, and Sam Altman sat on chip strategy, control vs. equity, timescales, perception of risk, the Q3 funding freeze, and the Cerebras merger.\n\n## Document type\n**Email thread, plain text, three messages stitched.** Zilis's August 21, 2017 6:05:17 PM forward (\"FYI!\") on top of Sam Teller's August 20, 2017 11:53 PM reply, on top of Zilis's same-day 11:40 PM original. Recipients of the forward: Jared Birchall (`jbirchall@muskfoundation.org`). Subject \"Fwd: OpenAI possible notes.\" Bates OPENAI_MUSK00023469–23470. Document ID DX-0651.\n\n## Logistics\n- **Trial admission:** Day 8 (May 6, 2026) — used during Zilis live direct. No prior wiki reference; admission day inferred from Box upload date.\n- **Box upload:** 2026-05-06 14:57:06 PT — Day 8 mid-afternoon defense batch.\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~434 KB, 2 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `0651.pdf`.\n\n## Transcribed text\n\n> **From:** Shivon Zilis <shivon@openai.com>\n> **To:** Jared Birchall <jbirchall@muskfoundation.org>\n> **Subject:** Fwd: OpenAI possible notes\n> **Date:** 8/21/2017 6:05:17 PM\n>\n> FYI!\n>\n> Begin forwarded message:\n\n> **From:** Sam Teller <steller@openai.com>\n> **To:** Shivon Zilis <shivon@openai.com>\n> **Subject:** Re: OpenAI possible notes\n> **Date:** August 20, 2017 at 11:53:26 PM PDT\n>\n> Thank you. TBH I think unless he's going to have a big call with Greg tomorrow you can communicate this stuff over the phone to E and in person Tues.\n\n> **From:** Shivon Zilis <shivon@openai.com>\n> **To:** Sam Teller <steller@openai.com>\n> **Subject:** OpenAI possible notes\n> **Date:** Aug 20, 2017, at 11:40 PM\n>\n> Unsure if any helpful since it's mostly soft humany things. Lmk what you think, steller!\n>\n> **Chip strategy:** One thing that Greg and Ilya were definitely not factoring into their mental model was the degree to which you're committed to winning the AI hardware battle. They don't know about the conversation you had with Jim re: creating a fab and putting real resources behind the effort. If they knew it, it would likely would deeply color their expectations for who ought to get what in the new structure.\n>\n> **Control vs. Equity:** Greg, Ilya, and Sam have wondered aloud about how you think about those two things and if there is any difference. Greg and Ilya seems less sensitive to short-term control than they are long-term control and having a solid amount of equity. Sam seems less sensitive on equity and more sensitive on control but has been by far the most flexible of the bunch so far.\n>\n> **Timescales:** It feels like Greg, Ilya, and Sam could potentially get on board with 2-3 years of control but are not currently in a mindset to be ok with it for longer than that, given their expectations of how quickly AGI is likely to emerge.\n>\n> **Perception of risks:** Greg seems much more concerned about China than he is Demis or DeepMind. One of his concerns about control is if there is a difference of opinions on strategy because of this divergent beliefs, he will not be able to argue his point. I, of course, did not want to speak for you but shared that in everything I've ever observed, you have always been willing to listen to new data points and update your belief if it's warranted. May be worth reinforcing that.\n>\n> **Funding freeze:** OpenAI is likely to realize this week that their $5M in Q3 funding is, albeit correctly, on hold. Unsure how this will impact negotiations but wanted to flag it since it's likely to have a big psychological impact on them if they find out.\n>\n> **Time commitment:** A point that keeps surfacing for them is how much time you would be able to commit and in what timeframe. They seems to align time and equity split very closely in their minds, whether rightly or wrongly. Giving some guidance on that when proposing a structure may be helpful.\n>\n> **Cerebras:** In the meeting this week with Cerebras, Sam A made good on his intention to reset Andrew's expectations to this being a longer process than he was initially hoping for (I'd had the same conversation with Andrew on Monday). Andrew made it clear to Sam that he 100% wanted to move forward. Sam's guess is that, outside of his desire for a deal, he is afraid that Michael James may leave to join OpenAI if the merger doesn't happen.\n\n## Commentary\n\nDX 651 is an extraordinary contemporaneous diagnostic of the **August 2017 negotiation**, written by Zilis from inside OpenAI to Musk's family-office circle. Defense uses three pieces of it: (i) **\"Greg and Ilya seems less sensitive to short-term control than they are long-term control and having a solid amount of equity\"** — the founders' true reservation price was *long-term governance + equity*, not absolute opposition to a Musk-led for-profit, which defense pairs with [[PX 157]]'s \"**we are happy to give up on the equity, personal control, make ourselves easily firable**\" to argue the founders were *flexible* and Musk was the rigid party; (ii) **\"Sam seems less sensitive on equity and more sensitive on control but has been by far the most flexible of the bunch so far\"** — defense reads this as evidence that Altman, contrary to plaintiffs' \"Altman the manipulator\" framing, was the moderating force throughout; and (iii) the **funding freeze** flag — the $5M Q3 hold — is the predicate for [[PX 157]]'s \"**I will no longer fund OpenAI**\" line; Musk had already been freezing money. The chip-strategy passage references Musk's conversations with **Jim Keller** about a fab — a thread that surfaces again in [[DX 773]] (Keller-Karpathy June 2017 thread). Cross-reference: [[DX 627]] (Zilis's prior briefing one month earlier) and [[DX 712]] (the October 1, 2017 follow-up structure memo).\n\n---\n*See also:* [[Day 8|Day 8 digest]] · [[Shivon Zilis]] · [[Sam Teller]] · [[Jared Birchall]] · [[DX 627]] · [[DX 712]] · [[PX 157]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-653", "exhibit": "DX 653", "party": "Defendants", "type": "Text messages", "admitted_trial_day": "Day 7 (May 5, 2026)", "uploaded_box_pt": "2026-05-05T14:26:59", "uploader": "Morrison Foerster", "pages": 2, "size_bytes": 129127, "source_pdf": "DX-653.pdf", "pdf_url": "https://media.mts-in.com/DX-653.pdf", "body_markdown": "# DX 653 — Aug 22, 2017 Sutskever text messages: \"At least we're getting our Teslas! Will a model 3 make you be willing to accept massively unfavorable terms?\"\n\n> Two-message text exchange from Ilya Sutskever, in a four-participant chat including Greg Brockman, sent at the high point of the August 2017 control negotiation — Sutskever joking that the Founder Series Tesla Model 3 gifts (see [[DX 631]]) might be Musk's way of softening the founders for \"massively unfavorable terms.\"\n\n## Document type\n**Text messages.** A \"Short Message Report\" extract showing one conversation, two messages, four participants, on phone number +1 408-332-1092 dated 8/22/2017. Identified participants include Greg Brockman, Ilya Sutskever, and \"Me\" (the producing custodian — Brockman). Bates OPENAI_MUSK00023473–23474.\n\n## Logistics\n- **Trial admission:** Day 7 (May 5, 2026) — Brockman defense direct + Mr. Molo recross + DX 1252; Wu deposition. Used by defense in connection with the August 2017 control-negotiation timeline — Sutskever's contemporaneous joke is the live-time companion to Brockman's PX 151 / PX 154 / PX 161 journal passages and to [[DX 631]] (the July 21, 2017 Founder Series Tesla gifts).\n- **Box upload:** 2026-05-05 14:26:59 PT — Day 7 mid-afternoon batch.\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~129 KB, 2 pages (cover + message log).\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `0653.pdf`.\n\n## Transcribed text\n\n> **Short Message Report**\n> Conversations: 1 · Total Messages: 2 · Participants: 4 · Date Range: 8/22/2017\n>\n> +1 408-332-1092 · 2 messages on 8/22/2017\n> Participants: Greg Brockman, Ilya Sutskever, Me (Greg Brockman) [+ one redacted]\n>\n> ---\n>\n> **Ilya Sutskever** — 8/22/2017, 10:13 PM\n> At least we're getting our Teslas!\n>\n> **Ilya Sutskever** — 8/22/2017, 10:15 PM\n> Will a model 3 make you be willing to accept massively unfavorable terms?\n\n## Commentary\n\nDX 653 is a small, surgical piece of contemporaneous color from the height of the August 2017 control negotiation. **Defense's read:** Sutskever's text is a joke, but the setup — Musk's Founder Series Tesla Model 3 gifts (see [[DX 631]]) coming through *while* Musk is asking for 51% control of the for-profit (see [[PX 156]] and [[PX 157]]) — paints Musk as someone who mixes patronage with hard-edged term demands. **Plaintiffs' read:** Sutskever's text — \"**Will a model 3 make you be willing to accept massively unfavorable terms?**\" — is a tell that the founders themselves understood Musk's August 2017 demands as massively unfavorable, i.e., not the kind of terms a person would accept absent gifts and pressure. The August 22, 2017 timestamp puts the exchange one day after Brockman's PX 151 journal entry (\"This is the only chance we have to get out from under Elon … take me to $1,000,000,000,\" see [[Key Themes]] § \"The Brockman journal\") and roughly a month before the [[PX 157]] \"final straw\" email of September 20. The redacted fourth participant is most likely Sam Altman or Wojciech Zaremba.\n\n---\n*See also:* [[Day 7|Day 7 digest]] · [[DX 631]] · [[PX 151]] · [[PX 156]] · [[PX 157]] · [[Ilya Sutskever]] · [[Greg Brockman]] · [[Brockman Journal]]\n"}
{"exhibit_id": "DX-657", "exhibit": "DX 657", "party": "Defense (OpenAI/MS)", "type": "Email", "admitted_trial_day": "Day 8 (May 6, 2026)", "uploaded_box_pt": "2026-05-06T14:57:06", "uploader": "Morrison Foerster", "pages": 1, "size_bytes": 420038, "source_pdf": "DX-657.pdf", "pdf_url": "https://media.mts-in.com/DX-657.pdf", "body_markdown": "# DX 657 — Aug 27, 2017 Zilis \"OpenAI Notes from Today\" memo to Teller & Birchall\n\n> Shivon Zilis's Sunday-night recap to Sam Teller and Jared Birchall after a \"supposed-to-be-half-hour\" meeting with Greg Brockman that \"turned into a two hour meeting with Greg and Ilya.\" Seven numbered questions — Amount of Control, Duration of Control, Time Spent, What to do with time spent, Ratio of time spent to control, Equity Split, Capitalization Strategy — written three weeks before the [[PX 157]] \"final straw\" rupture. The defense's clearest contemporaneous evidence that the founders' \"**ironclad agreement to not have Elon (or anyone) have absolutely control**\" was on the table by **late August 2017**.\n\n## Document type\n**Email.** Plain-text email from Shivon Zilis <shivon@neuralink.com> to Sam Teller <steller@spacex.com> and Jared Birchall <jared@excession.com>, Sun, 27 Aug 2017 19:19:03 -0700, subject \"OpenAI Notes from Today.\" Bates Neuralink_0000744. Production-stamped Highly Confidential-AEO.\n\n## Logistics\n- **Trial admission:** Day 8 (May 6, 2026) — Murati live; Zilis live; Toner video begins. Used by Schubert (defense) on Zilis cross alongside [[DX-758]] (the Feb 2018 nine-scenarios memo) — see [[Key Themes#\"Unequivocal control\" (the 2017 negotiation breakdown)|\"Unequivocal control\"]].\n- **Box upload:** 2026-05-06 14:57:06 PT — Day 8 mid-afternoon batch (~8 seconds after [[DX-758]]).\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~410 KB, 1 page.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `0657.pdf`.\n\n## Transcribed text\n\n> **From:** Shivon Zilis <shivon@neuralink.com>\n> **To:** Sam Teller <steller@spacex.com>, Jared Birchall <jared@excession.com>\n> **Subject:** OpenAI Notes from Today\n> **Date:** Sun, 27 Aug 2017 19:19:03 -0700\n>\n> This was supposed to be a half hour meeting with Greg that turned into a two hour meeting with Greg and Ilya. This is the distillation of the random walk we took… 7 general outstanding questions and their commentary below.\n>\n> **1. Amount of Control?**\n> -Does Elon require absolute control? Is there some sort of scenario where he would accept some sort of creative overrule provision if everyone else (not just the three of them, but perhaps a board including others) disagreed on direction?\n>\n> **2. Duration of Control?**\n> -A complete non-negotiable for all three of them is an ironclad agreement to not have Elon (or anyone) have absolutely control of AGI they create. Satisfying this for them means a situation where, regardless of what happens to the three of them, it's guaranteed that power over the company is distributed after the 2-3 year timeframe.\n>\n> **3. Time Spent?**\n> -How much time does Elon want to spend on this? How much time can Elon \\*actually\\* spend on this? In what timeframe? They are comfortable with more control if more time spent. Their fear is that if he's only spending an hour a week he will not have all of the contextual information he needs to make the right decisions. Note that literally everyone acknowledges he has the most accurate visionary sense of anyone they have ever met by orders of magnitude — they just want to ensure that he has committed to a structure where he will consistently have sufficient forced contextual awareness.\n>\n> **4. What to do with time spent?**\n> -They don't really know how Elon spends his time at the other companies and what he would want to do with OpenAI. I was able to shed light on the former, but didn't want to speak to the latter. If Elon did spend 5+ hours a week committed to this, what would he want that to look like?\n> -Greg and Ilya are confident they could build out SW / algos. They are not confident on making the right hardware choices and building out those teams. If Elon could make that happen with his hours committed they think it would be a really amazing situation.\n>\n> **5. Ratio of time spent to control?**\n> -They are not ok with low time spent / high control. They are cool with less time / less control.\n>\n> **6. Equity Split?**\n> -Greg still super stuck on this out of an instinct for fairness. I personally disagree with him on many counts for this and shared the others things I thought he should fit into his mental model (Elon's ability to create amazing AGI hardware, raise future money at insanely good rates, accelerate the speed of development, recruit kickass people, Elon's contributions so far).\n> -He did share a few helpful things though… including that Ilya in some ways has contributed millions by leaving his earning potential on the table at Google, that Greg is literally committing his entire life to this (keep between us but just broke up with this unbelievably awesome woman I know literally because he wanted to focus all his time to safe AGI and he told her that).\n> -Note: one concern they had was his proposed employee pool was too small. Would have to factor that in.\n>\n> **7. Capitalization Strategy?**\n> -$100M and change doesn't make sense to these guys. They are of the opinion that the datacenter they need alone would cost that. They are guessing the best thing to do is raise much more out of the gate.\n>\n> **Takeaways:**\n> Unsure if any of this would work on Elon's end but a structure that would likely \\*just\\* work for them is:\n> -Elon either spend 5-10 hours a week and has near full control, or spend less time and has less control\n> -There is a creative short-term override in a completely extreme scenario that was not just Greg / Sam / Ilya.\n> -There is an iron clad 2-3yr minority control agreement, regardless of the fates of Greg / Sam / Ilya\n> -There is 200M-1B raised initially.\n> -Greg and Ilya stakes end up higher that 1/10 of Elon's, but not significantly so.\n\n## Commentary\n\nDX 657 is the August-27, 2017 minutes that — read against [[PX 157]] (Sept. 20, 2017) — let the defense argue the \"**absolute control**\" theme was diagnosed by Brockman and Sutskever *to Zilis* a full three weeks before the \"Honest Thoughts\" letter ever went out. Question 1 (\"**Does Elon require absolute control?**\") and Question 2 (\"**ironclad agreement to not have Elon (or anyone) have absolutely control of AGI they create**\") map directly onto the [[PX 157]] line \"**absolute control is extremely important to you**.\" The exhibit also seeds the \"**Greg / Sam / Ilya**\" ownership split that Brockman and Birchall would later be cross-examined on — see [[Key Themes#Greg Brockman's contributions (\"zero\")|Brockman's contributions]] and the **$30B / $0** trade documented on Day 6. The flattering \"literally everyone acknowledges he has the most accurate visionary sense of anyone they have ever met by orders of magnitude\" line is also defense gold for the \"Musk-was-not-betrayed-he-was-resisted\" frame: the founders weren't dismissive of Musk; they were trying to box his control, not his vision. Pair with [[DX-758]] (Feb 2018 \"OpenAI / Possible AI scenarios\").\n\n---\n*See also:* [[Day 8|Day 8 digest]] · [[Shivon Zilis]] · [[PX 157]] · [[DX-758]] · [[Greg Brockman]] · [[Ilya Sutskever]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-662", "exhibit": "DX 662", "party": "Defense", "type": "Email thread", "admitted_trial_day": "Day 3 (April 29, 2026)", "uploaded_box_pt": "2026-04-29T15:37:43", "uploader": "Morrison Foerster", "pages": 1, "size_bytes": 357303, "source_pdf": "DX-662.pdf", "pdf_url": "https://media.mts-in.com/DX-662.pdf", "body_markdown": "# DX 662 — Aug 28–29, 2017 Chris Clark / Birchall / Musk thread on the Q3 2017 OpenAI grant (\"compute payment covering what we used for the Dota project\")\n\n> Chris Clark's Aug 28, 2017 email to Birchall asking when Musk's quarterly **$5 million** OpenAI grant would land — the compute bill for \"the Dota project (the one that got all the press a couple weeks ago)\" was due in September — and Musk's one-word \"**Yes**\" approving Birchall's plan to **continue holding** the funding.\n\n## Document type\n**Email thread, plain text, three messages.** Chris Clark (chris@openai.com) → Jared Birchall, Mon Aug 28, 2017 12:32 PM, \"Q3 2017 Grant\"; Birchall forward to Musk, Aug 29, 2017 12:07 AM, \"For now I have held off on the quarterly $5M. Continue to hold?\"; Musk reply Aug 29, 2017 07:18:08 -0000, \"Yes.\" Bates 2024MUSK-0009625.\n\n## Logistics\n- **Trial admission:** Day 3 (April 29, 2026) — Musk direct concludes; cross begins. Used by defense to establish that the August 2017 funding pause was deliberate and predated the [[PX 157]] \"final straw\" email by ~3 weeks — and that **Musk's funding leverage was the live mechanism** behind the September 2017 control fight.\n- **Box upload:** 2026-04-29 15:37:43 PT — Day 3 late-afternoon Defense batch (clustered with [[DX 819]], [[DX 844]], [[DX 1444]]).\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~349 KB, 1 page.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `662 .pdf` (note trailing space in filename).\n\n## Transcribed text\n\n> **From:** \"Elon Musk\" <erm@spacex.com>\n> **To:** \"Jared Birchall\" <jbirchall@muskfoundation.org>\n> **Subject:** Re: Q3 2017 Grant\n> **Date:** Tue, 29 Aug 2017 07:18:08 -0000\n> **Importance:** Normal\n>\n> Yes\n>\n> On Aug 29, 2017, at 12:07 AM, Jared Birchall <jbirchall@muskfoundation.org> wrote:\n>\n> For now I have held off on the quarterly $5M. Continue to hold?\n>\n> ---------- Forwarded message ----------\n> **From:** Chris Clark <chris@openai.com>\n> **To:** Jared Birchall <jbirchall@muskfoundation.org>\n> **Subject:** Q3 2017 Grant\n> **Date:** Mon, Aug 28, 2017 at 12:32 PM\n>\n> Hi Jared,\n>\n> I just wanted to check in re: Q3 2017 grant. **The compute payment covering what we used for the Dota project (the one that got all the press a couple weeks ago) will likely be due in Sept, so I wanted to double check on timing.**\n>\n> Thanks,\n> Chris\n\n## Commentary\n\nDX 662 is defense's **smoking gun for the \"Musk's funding was the leverage\" theory**: as of August 28, 2017 — three weeks before the [[PX 157]] \"final straw\" email and one week after the [[PX 151]] / Aug 21 \"this is the only chance we have to get out from Elon\" Brockman journal entry — OpenAI had a **compute bill due in September** for the Dota project, the **$5M quarterly funding had been frozen**, and Musk's one-word \"**Yes**\" continued the freeze. This is the contemporaneous corroboration of Zilis's August 2017 internal memo (testified to on Day 8) that \"**As of August 20, Mr. Musk hadn't yet told OpenAI he'd frozen their funding**\" (see [[Key Themes]] §\"the 'ironclad agreement' red line — and the funding freeze\"). Plaintiffs use the same fact two ways: (i) Musk had frozen OpenAI's funding before the founders' Aug 28 Zilis memo (PX 152) reached Musk's inbox at 12:08 AM, which means the founders were already operating under financial duress when they proposed the \"ironclad 2-3 year minority control agreement\"; (ii) OpenAI nevertheless honored its commitments and Musk's later \"I won't fund OpenAI until you make a firm commitment to stay\" line in PX 157 was an after-the-fact rationalization.\n\n---\n*See also:* [[Day 3|Day 3 digest]] · [[PX 151]] · [[PX 152]] · [[PX 157]] · [[Brockman Journal]] · [[Jared Birchall]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-664", "exhibit": "DX 664", "party": "Defense", "type": "Memo", "admitted_trial_day": "Day 8 (May 6, 2026)", "uploaded_box_pt": "2026-05-06T14:57:07", "uploader": "Morrison Foerster", "pages": 1, "size_bytes": 363326, "source_pdf": "DX-664.pdf", "pdf_url": "https://media.mts-in.com/DX-664.pdf", "body_markdown": "# DX 664 — Aug 30, 2017 Zilis → Teller, \"OpenAI starting sketch\" (B Corp / C Corp options)\n\n> A one-page Shivon Zilis memo to Sam Teller on August 30, 2017 — an \"OpenAI starting sketch\" for the for-profit conversion: founder slate (Musk Executive Chairman/CEO?), $135M initial raise (Elon $110M / 55%, Altman $15M / 7.5%), 17.5% employee pool, and two operating-structure options: \"**Option 1: Roll everything into a B Corp (would include all of OpenAI)**\" / \"**Option 2: OpenAI C Corp and OpenAI non-profit**.\"\n\n## Document type\n**Internal memo (sent as email body), plain text.** From Shivon Zilis <shivon@neuralink.com> to Sam Teller <sam@neuralink.com>, sent Wed, 30 Aug 2017 14:52:55 -0700, subject \"OpenAI starting sketch.\" Bates Neuralink_0000729. Marked Highly Confidential-AEO.\n\n## Logistics\n- **Trial admission:** Day 8 (May 6, 2026 — Murati live; Zilis live; Toner video begins). Per the Day 8 quotes file: \"**DX 664** — Zilis 2017 notes to Sam Altman: 'Option one, **roll everything into a B Corp**. … Option two, **OpenAI C Corp and OpenAI nonprofit**.'\" Used during Zilis live testimony.\n- **Box upload:** 2026-05-06 14:57:07 PT — Day 8 mid-afternoon defense batch (clustered with DX 619, DX 724, DX 757, DX 824, DX 900).\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~355 KB, 1 page.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `0664.pdf`.\n\n## Transcribed text\n\n> **From:** Shivon Zilis <shivon@neuralink.com>\n> **To:** Sam Teller <sam@neuralink.com>\n> **Subject:** OpenAI starting sketch\n> **Date:** Wed, 30 Aug 2017 14:52:55 -0700\n>\n> What else do we want to cover / edit?\n>\n> **Founders:**\n> - Elon Musk (Executive Chairman? CEO?)\n> - Greg Brockman (Founder, CTO)\n> - Ilya Sutskever (Founder, Head of Artificial Intelligence)\n> - Sam Altman (CEO, Chairman, none of the above?)\n>\n> **Governing Structure:**\n> - One vote per share.\n> - Board of 8-10.\n> - Overrule provision where *unanimous* agreement from the rest of the board.\n> - Plan to move to minority stake / voting rights within 1-2 successive round of financing.\n>\n> **Capitalization Structure:**\n> - $135M initial raise\n>   - Elon: $110M (55%)\n>   - Sam Altman: $15M (7.5%)\n>   - Other investors: $10M (5%)\n> - Founder shares\n>   - Greg Brockman: 15M (7.5%)\n>   - Ilya Sutskever: 15M (7.5%)\n>   - Employee Pool: 30M (17.5%)\n>\n> **Operating Structure:**\n> - Option 1: Roll everything into a B Corp (would include all of OpenAI)\n> - Option 2: OpenAI C Corp and OpenAI non-profit\n\n## Commentary\n\nDX 664 is defense's **\"the founders, including Zilis, were openly modeling B Corp / C Corp conversions in August 2017\"** receipt — i.e., the for-profit conversion was a contemporaneous, signaled-to-Musk strategic option, not a covert later betrayal. The memo's two options track the same architecture Brockman recorded in his **Sept 12, 2017** journal entry (see [[Brockman Journal|the PX 154 phone-call notes]]): \"*do we transition from nonprofit to something which is essentially philanthropic endeavor and is B Corp or C Corp or something*\"; \"*Elon need to understand B Corp situation*\"; \"*Elon agrees [B Corp] does sound better*.\" DX 664 is the **inside-Musk-team** version of that discussion two weeks earlier — sent between Zilis (then transitioning to OpenAI's board) and Teller (Musk's chief of staff) — showing the founders considered Option 2 (a C Corp and a parallel nonprofit) as a real structural choice. The proposed cap-table — **Musk 55%, Altman 7.5%, Brockman/Sutskever 7.5% each, employee pool 17.5%** — sits between the August 18 [[PX-1250-16|\"definitely not unilateral control … equal split\"]] entry and the September 11 [[PX 153|cap-table proposal]] (Musk 51.20%, founders ~11% each). The architecture that actually shipped — the 2018 **OpenAI LP capped-profit** plus the **OpenAI Inc. nonprofit parent** — is essentially **Option 2** of this memo. Used by defense to show OpenAI's evolution into a for-profit-with-nonprofit-parent was a 2017 **founders-and-Zilis-agreed-upon** option, not a 2018 surprise. Cross-reference [[Key Themes]] § \"Day 8 — what Zilis added.\"\n\n---\n*See also:* [[Day 8|Day 8 digest]] · [[Shivon Zilis]] · [[Brockman Journal]] · [[PX 153]] · [[DX-827|DX 827]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-668", "exhibit": "DX 668", "party": "Defense", "type": "Email thread", "admitted_trial_day": "Day 4 (April 30, 2026)", "uploaded_box_pt": "2026-04-30T15:13:13", "uploader": "Morrison Foerster", "pages": 1, "size_bytes": 368229, "source_pdf": "DX-668.pdf", "pdf_url": "https://media.mts-in.com/DX-668.pdf", "body_markdown": "# DX 668 — Sept 1, 2017 Clark/Birchall \"Q3 2017 Grant\" — Musk's funding pause confirmed in writing\n\n> Three-message email thread between Chris Clark (OpenAI) and Jared Birchall (Musk Foundation), Aug 28 – Sept 1, 2017, in which Birchall tells Clark \"**Elon informed Greg and Ilya that the funding would be on pause until they came to terms on the right path moving forward**\" and Clark responds he \"**will do what I can to reduce or delay OpenAI expenditures until the concerns are resolved**.\"\n\n## Document type\n**Email thread, single page, three messages.** Production-stamped Confidential; Bates EXMF-0003122. Document ID DX-0668. Earliest message Mon, 28 Aug 2017 12:32 PM (Clark → Birchall, \"Q3 2017 Grant\"); Birchall reply Thu, 31 Aug 2017 11:52 PM; Clark reply Fri, 1 Sep 2017 7:09:17 AM.\n\n## Logistics\n- **Trial admission:** Day 4 (April 30, 2026) — Musk cross + redirect; Birchall direct + cross. Used as a contemporaneous-record exhibit during the Birchall block to corroborate the timing and content of Musk's August 2017 funding freeze on OpenAI — the squeeze that culminated in the [[PX 157]] \"**final straw**\" exchange on September 20.\n- **Box upload:** 2026-04-30 15:13:13 PT — Day 4 mid-afternoon batch (clustered with DX 621 and DX 1256).\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~360 KB, 1 page.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `668.pdf`.\n\n## Transcribed text\n\n> **From:** Chris Clark <chris@openai.com>\n> **To:** Jared Birchall <jbirchall@muskfoundation.org>\n> **Subject:** Re: Q3 2017 Grant\n> **Date:** 9/1/2017 7:09:17 AM\n>\n> Thanks. They shared the same info with me today as well. I will do what I can to reduce or delay OpenAI expenditures until the concerns are resolved.\n>\n> FYI, you will likely see a few joint requests soon from Neuralink/OpenAI re: security, HVAC, and other expenses we hope won't be controversial and probably shouldn't wait too long. They benefit both companies equally and have been requested by Elon's team during visits (better climate control in specific conference rooms shared by both orgs, 24/7 security, and better access controls).\n>\n> I'll send you the proposals and estimates, and then I'll defer to you about where or not we move forward with them.\n>\n> Chris\n\n> **From:** Jared Birchall <jbirchall@muskfoundation.org>\n> **To:** Chris Clark <chris@openai.com>\n> **Subject:** Re: Q3 2017 Grant\n> **Date:** Thu, Aug 31, 2017 at 11:52 PM\n>\n> Hey Chris,\n>\n> This was ready to go out when I was told that Elon informed Greg and Ilya that the funding would be on pause until they came to terms on the right path moving forward.\n>\n> Jared\n\n> **From:** Chris Clark <chris@openai.com>\n> **To:** Jared Birchall <jbirchall@muskfoundation.org>\n> **Subject:** Q3 2017 Grant\n> **Date:** Mon, Aug 28, 2017 at 12:32 PM\n>\n> Hi Jared,\n>\n> I just wanted to check in re: Q3 2017 grant. The compute payment covering what we used for the Dota project (the one that got all the press a couple weeks ago) will likely be due in Sept, so I wanted to double check on timing.\n>\n> Thanks,\n> Chris\n\n## Commentary\n\nDX 668 is the contemporaneous **paper trail** for the August 2017 Musk funding freeze that Zilis described from her own files on Day 8 (\"**As of August 20, Mr. Musk hadn't yet told OpenAI he'd frozen their funding**\" — `5/6/2026 Testimony @ ~11:25 PT`; see [[Key Themes]]). What Zilis observed informally from inside OpenAI in late August is here documented in writing between the **Musk Foundation's CFO** (Birchall) and **OpenAI's CFO** (Clark): Q3 grant on hold, pending \"the right path moving forward.\" The exhibit cuts both ways. For plaintiffs, it shows Musk's funding *was* the lever — which supports the \"I'll no longer fund OpenAI until you have made a firm commitment to stay\" line in [[PX 157]] as a real, in-flight ultimatum tying donations to control of governance. For defense, it is the same fact pattern that supports the [[Key Themes#\"Pageant of hypocrisy\" / Tale of two Elons|\"picked up his marbles\"]] / unequivocal-control framing — Musk used dollars as a control instrument while the founders were resolving they could not let him have unilateral AGI authority (see [[PX 151]] and [[Brockman Journal]]). Note also Clark's deference in the second paragraph — Birchall, not OpenAI's board, decides which Pioneer Building expenses Neuralink/OpenAI may share — which dovetails into the [[Key Themes#The Pioneer Building|Pioneer Building]] thread where Birchall told Sam Teller \"Elon holds 100 percent of the power and authority regarding the building\" (DX 600).\n\n---\n*See also:* [[Day 4|Day 4 digest]] · [[Jared Birchall]] · [[Chris Clark]] · [[PX 151]] · [[PX 157]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-669", "exhibit": "DX 669", "party": "Defense", "type": "Text messages", "admitted_trial_day": "Day 8 (May 6, 2026)", "uploaded_box_pt": "2026-05-06T14:57:08", "uploader": "Morrison Foerster", "pages": 3, "size_bytes": 357600, "source_pdf": "DX-669.pdf", "pdf_url": "https://media.mts-in.com/DX-669.pdf", "body_markdown": "# DX-669 — Sept 1, 2017 Teller/Zilis texts: \"stopping OpenAI payments for now\"\n\n> Sam Teller asks Shivon Zilis whether Musk told Greg and Ilya he was halting OpenAI funding; Zilis confirms — three weeks before [[PX 157]] and the \"I will no longer fund OpenAI\" line.\n\n## Document type\n**Text messages** (Apple Short Message Report extract). One conversation, 32 messages on 9/1/2017, two participants: Sam Teller and Shivon Zilis. Bates ZILIS-0000932–0000934.\n\n## Logistics\n- **Trial admission:** Day 8 (May 6, 2026) — Murati live; Zilis live; Toner video begins.\n- **Box upload:** 2026-05-06 14:57:08 PT — Day-8 batch.\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~358 KB, 3 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `0669.pdf`.\n\n## Transcribed text\n\n> **Short Message Report**\n> Conversations: 1 · Total Messages: 32 · Date Range: 9/1/2017\n> Participants: Sam Teller, Shivon Zilis\n>\n> **Sam Teller** [12:21 AM]: Did Elon say to Greg and Ilya that he was stopping OpenAI payments for now?\n> **Shivon Zilis** [12:21 AM]: Yes\n> **Shivon Zilis** [12:22 AM]: Right after \"I decline\"\n> **Shivon Zilis** [12:22 AM]: He asked when they were going to leave. They said they weren't planning to, and he said until resolved funding held\n> **Shivon Zilis** [1:07 AM]: Hey! Not sure when you head to Ecuador but if anything I can keep an eye out for next week for you lmk!\n> **Sam Teller** [1:28 AM]: We can talk tomorrow\n> **Shivon Zilis** [8:48 PM]: You have anything between 3 and 5?\n> **Sam Teller** [8:48 PM]: Nah\n> **Shivon Zilis** [8:49 PM]: K cool. Just email me or call me on weekend if you need to give me any heads up on stuff I can cover for you next week\n> **Sam Teller** [8:55 PM]: Most concerned about OpenAI sitch\n> **Sam Teller** [8:55 PM]: E may go to Australia Sun-Weds\n> **Sam Teller** [8:55 PM]: So will have to be thoughtful about setting up calls etc\n> **Shivon Zilis** [8:56 PM]: Ah\n> **Shivon Zilis** [8:56 PM]: Shoot, will be tricky if he's not here two weeks in a row on that and NL front\n> **Sam Teller** [8:57 PM]: Yes\n> **Shivon Zilis** [8:57 PM]: We should figure out how to best manage. I will think on it\n> **Sam Teller** [8:57 PM]: Could be tough patch\n\n## Commentary\n\nDX 669 is the contemporaneous insider record of how Musk's late-August 2017 \"I decline\" landed inside his own organization. Zilis confirms Musk told Brockman and Sutskever that **\"funding held\"** until the for-profit conversion was resolved — three weeks ahead of the [[PX 157]] \"I will no longer fund OpenAI\" reply. The defense uses this on [[Day 8|Day 8]] (Zilis live) to show Musk was explicitly leveraging payments against the negotiation, not merely \"pausing\"; it also feeds the [[Key Themes]] line that Musk treated his donations as conditional on control. Sam Teller's worried \"**Most concerned about OpenAI sitch**\" is the same internal anxiety that runs through Brockman's Aug. 21, 2017 [[PX 151]] entry and the Sept. 20 [[PX 157]] \"Honest Thoughts\" letter.\n\n---\n\n*See also:* [[Day 8|Day 8 digest]] · [[Shivon Zilis]] · [[PX 157]] · [[PX 151]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-679", "exhibit": "DX 679", "party": "Defense", "type": "Email", "admitted_trial_day": "Day 3 (April 29, 2026)", "uploaded_box_pt": "2026-04-29T15:36:42", "uploader": "Morrison Foerster", "pages": 1, "size_bytes": 473721, "source_pdf": "DX-679.pdf", "pdf_url": "https://media.mts-in.com/DX-679.pdf", "body_markdown": "# DX 679 — Sept 11, 2017 Musk → Brockman, \"you are pushing too hard here\"\n\n> Musk's one-line reply (\"**Guys, you are pushing too hard here. I'm not ok with this.**\") on top of Brockman's September 10, 2017 proposed cap-table spreadsheet — the founders' mid-September 2017 attempt to share equity that lit the fuse for the [[PX 157]] \"final straw\" email nine days later.\n\n## Document type\n**Email thread, plain text, two messages.** Musk's September 11, 2017 10:56 PM PDT reply on top of Brockman's September 10, 2017 8:51 PM email proposing a cap table via Google Sheets link. Recipients: Greg Brockman, cc Sam Altman <shga@ycombinator.com>, Ilya Sutskever. Subject \"Re: proposed cap table.\" Bates 2024MUSK-0009373. Document ID DX-0679.\n\n## Logistics\n- **Trial admission:** Day 3 (April 29, 2026) — used by defense during Musk cross to anchor the September 2017 control fight as a documented Musk-side rejection of cap-table sharing. No prior wiki reference; admission day inferred from Box upload date.\n- **Box upload:** 2026-04-29 15:36:42 PT — Day 3 late-afternoon batch (clustered with DX 827, DX 862, DX 927).\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~463 KB, 1 page.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `679 .pdf` (note trailing space in filename).\n\n## Transcribed text\n\n> **From:** Elon Musk <erm@spacex.com>\n> **To:** Greg Brockman <gdb@openai.com>\n> **Cc:** Sam Altman <shga@ycombinator.com>, Ilya Sutskever <ilyasu@openai.com>\n> **Subject:** Re: proposed cap table\n> **Date:** Mon, 11 Sep 2017 22:56:26 -0000\n>\n> **Guys, you are pushing too hard here. I'm not ok with this.**\n\n> **From:** Greg Brockman <gdb@openai.com>\n> **To:** Elon Musk <erm@spacex.com>\n> **Cc:** Sam Altman <shga@ycombinator.com>, Ilya Sutskever <ilyasu@openai.com>\n> **Subject:** proposed cap table\n> **Date:** Sep 10, 2017, at 8:51 PM\n>\n> Here's Ilya and my proposed cap table:\n> https://docs.google.com/spreadsheets/d/1zaEXQIqoUa_I7dfx-Omxm2cmxb0FiKNJ492YAj6dsEo/edit#gid=0\n>\n> Notes:\n>\n> - Sam and I are able to invest a lot more, and would be happy to trade off founder grants for increased investment if the current setup is an issue.\n> - The relative ranking of employees isn't final but is probably close (and people we don't want to bring aren't listed). We were generous with the amounts as they took a risk to join and have helped get us this far.\n> - As a reminder, to go above $2.5M in investment, Ilya would need to take a loan from Sam/me securitized by his stock. And investment dollars I generate by selling Stripe shares will be QSBS rollover eligible, if we structure things appropriately — though understood it may not be worth the complication.\n>\n> Please let us know what you think!\n>\n> - gdb\n\n## Commentary\n\nDX 679 is the connective tissue between [[DX 646]] (Aug 16) and [[PX 157]] (Sept 20). Brockman is doing exactly what plaintiffs claim Musk wanted in DX 646 — building the proposed for-profit cap table — and Musk responds with **\"you are pushing too hard here.\"** Defense uses this whiplash to argue that Musk had been the architect of the conversion idea but wouldn't accept *any* split that he didn't dominate. Two technical details on the page deepen the picture: (i) Brockman writes that he and Altman \"are able to invest a lot more, and would be happy to trade off founder grants for increased investment\" — i.e., to give up their own equity to make the math work — which Sutskever's \"**happy to give up on the equity, personal control, make ourselves easily firable**\" line in [[PX 157]] then escalates; (ii) Ilya would have to take a personal loan from Sam/Greg to invest above $2.5M — illustrating the founders' actual financial exposure if the deal closed and how far it was from a \"Musk gave them everything\" picture. The Stripe-shares QSBS-rollover detail also confirms Brockman was selling personal stock to fund OpenAI, complicating any narrative that the founders were free-riders. Cross-reference the [[Brockman Journal]] entries ([[PX 151]], [[PX 154]], [[PX 161]]) from this same window.\n\n---\n*See also:* [[Day 3|Day 3 digest]] · [[PX 157]] · [[DX 646]] · [[Greg Brockman]] · [[Ilya Sutskever]] · [[Brockman Journal]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-686", "exhibit": "DX 686", "party": "Defense", "type": "Email thread", "admitted_trial_day": "Day 3 (April 29, 2026)", "uploaded_box_pt": "2026-04-29T15:36:42", "uploader": "Morrison Foerster", "pages": 2, "size_bytes": 801154, "source_pdf": "DX-686.pdf", "pdf_url": "https://media.mts-in.com/DX-686.pdf", "body_markdown": "# DX 686 — Sept 13, 2017 Musk → Sutskever/Brockman \"Re: Current State\" (Series A 4/3 board, \"unequivocally have initial control\")\n\n> Musk's September 13, 2017 reply to Ilya Sutskever's prior-night summary of for-profit equity and board terms — the email in which Musk states he would \"**unequivocally have initial control of the company**,\" that the Series A he funded as \"supermajority me\" should appoint **four** board seats (vs. three common-stock seats for Sutskever/Brockman/Altman), and that at a hypothetical 16-person board he wanted \"25% influence\" as his minimum comfort level.\n\n## Document type\n**Email thread, plain text, two messages.** Musk's September 13, 2017 7:40:05 AM (UTC) reply on top of Ilya Sutskever's September 12, 2017 11:44 PM original. Recipients: Sutskever; cc Greg Brockman (handles redacted). Subject \"Re: Current State.\" Bates 2024MUSK-0009240–9241. Document ID DX-0686.\n\n## Logistics\n- **Trial admission:** Day 3 (April 29, 2026). Pre-trial wiki reference: **\"DX 686 — Sept 2017 unequivocal-control email.\"** Used during plaintiffs' Musk direct/cross — the foundation of the defense's [[Key Themes#\"Unequivocal control\" (the 2017 negotiation breakdown)|\"unequivocal control\"]] cross theme: **\"So the idea is, you would have unequivocal control of this for-profit when it starts; right?\" \"Yes.\"** `(042926TT.txt:3763)`.\n- **Box upload:** 2026-04-29 15:36:42 PT — Day 3 late-afternoon defense batch (clustered with [[DX 749]], [[DX 773]], [[DX 863]]).\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~782 KB, 2 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `686 .pdf` (note trailing space).\n\n## Transcribed text\n\n> **From:** Elon Musk <erm@spacex.com>\n> **To:** Ilya Sutskever <ilyasu@openai.com>\n> **Cc:** Greg Brockman <gdb@openai.com>\n> **Subject:** Re: Current State\n> **Date:** Wed, 13 Sep 2017 07:40:05 -0000\n>\n> Sounds good. The three common stock seats (you, Greg and Sam) should be elected by common shareholders. They will de facto be yours, but not in the unlikely event that you lose the faith of a huge percentage of common stockholders over time or step away from the company by choice.\n>\n> I think that the Preferred A investment round (supermajority me) should have the right to appoint four (not three) seats. I would not expect to appoint them immediately, but, like I said I would unequivocally have initial control of the company, but this will change quickly. The rough target would be to get to a 12 person board (probably more like 16 if this board really ends up deciding the fate of the world) where each board member has a deep understanding of technology, at least a basic understanding of AI and strong & sensible morals.\n>\n> Apart from the Series A four and the Common three, there would likely be a board member with each new lead investor/ally. However the specific individual new board members can only be added if all but one existing board members agrees. Same for removing board members.\n>\n> There will also be independent board members we want to add who aren't associated with an investor. Same rules apply: requires all but one of existing directors to add or remove.\n>\n> I'm super tired and don't want to overcomplicate things, but this seems approx right. At the sixteen person board level, we would have 7/16 votes and I'd have a 25% influence, which is my min comfort level. That sounds about right to me. If everyone else we asked to join our board is truly against us, we should probably lose.\n>\n> As mentioned, my experience with boards (assuming they consist of good, smart people) is that they are rational and reasonable. There is basically never a real hardcore battle where an individual board vote is pivotal, so this is almost certainly (sure hope so) going to be a moot point.\n>\n> As a closing note, I've been really impressed with the quality of discussion with you guys on the equity and board stuff. I have a really good feeling about this.\n>\n> Lmk if above seems reasonable.\n>\n> Elon\n\n> **From:** Ilya Sutskever <ilyasu@openai.com>\n> **To:** Elon Musk <erm@spacex.com>\n> **Cc:** Greg Brockman <gdb@openai.com>\n> **Subject:** Current State\n> **Date:** Sep 12, 2017, at 11:44 PM\n>\n> Hi Elon,\n>\n> To summarize our understanding of the current state:\n>\n> **On equity:**\n>\n> Greg: 10M grant/10M investment\n> Sam: 10M grant/10M investment\n> Ilya: 12M grant/2.5M investment + 5.5M loan from Greg securitized by Ilya's YC vested stock, from his work at OpenAI\n>\n> **On control:**\n>\n> 3 board seats for Elon, 1 board seat each for Ilya/Greg/Sam. Plan to expand board over time through unanimous consent of current board, up to 12 people. Details TBD on tiebreak and whether the directors are statically allocated.\n>\n> **On credit:**\n>\n> Elon is extremely happy to help Greg and Ilya get credit for their work — and is very open to any ideas Greg and Ilya may have.\n>\n> Looking forward to the conclusions of your conversation with Sam tomorrow.\n>\n> Ilya\n\n## Commentary\n\nDX 686 is the defense's **single most important Musk email** in the case. Three phrases drive the cross examination: **\"Preferred A investment round (supermajority me),\"** **\"I would unequivocally have initial control of the company,\"** and **\"7/16 votes and I'd have a 25% influence, which is my min comfort level.\"** Together these are Savitt's textual proof that the September 2017 negotiation broke down because Musk insisted on being the controlling shareholder and CEO of a for-profit OpenAI — exactly what Sutskever and Brockman then accuse him of in [[PX 157]] (\"**you've shown to us that absolute control is extremely important to you**\"). The Sutskever bottom message is also independently significant: it documents the **$10M+/$10M+ equity grants and investments** the founders had been negotiating to take in the for-profit (with **Greg loaning Ilya $5.5M securitized against Ilya's YC stock**) — i.e., the founders were prepared to put their own capital into the for-profit Musk was structuring. Musk's \"**this is almost certainly going to be a moot point**\" closing line is the defense's irony hook. Cross-reference: PX 156 (the same negotiation thread plaintiffs read in differently); [[PX 157]] (the Sept. 20 \"final straw\" reply seven days later); [[PX 158]] (Altman's Sept. 21 \"i remain enthusiastic about the non-profit structure!\"); see [[Key Themes]] for Cohen's \"**51 percent of the company**\" cross.\n\n---\n*See also:* [[Day 3|Day 3 digest]] · [[Ilya Sutskever]] · [[Greg Brockman]] · [[PX 157]] · [[PX 158]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-691", "exhibit": "DX 691", "party": "Defendants", "type": "Text messages", "admitted_trial_day": "Day 3 (April 29, 2026)", "uploaded_box_pt": "2026-04-29T15:36:42", "uploader": "Morrison Foerster", "pages": 1, "size_bytes": 151470, "source_pdf": "DX-691.pdf", "pdf_url": "https://media.mts-in.com/DX-691.pdf", "body_markdown": "# DX 691 — Sept 13, 2017 Musk text to Sam Altman, Jared Birchall, and Shivon Zilis: \"We should get going on creating the OpenAI B corp, as I promised Greg and Ilya.\"\n\n> A single text message from Elon Musk, sent at 7:43 AM PDT on September 13, 2017 — one week before the [[PX 157]] \"final straw\" email — affirmatively stating that Musk himself **promised Greg Brockman and Ilya Sutskever** that he would create the OpenAI B corp.\n\n## Document type\n**Text messages.** A single SMS in a four-participant chat (SMS-MMS-Chats platform). Participants: Sam [Altman], Jared Birchall (Musk's family-office head and OpenAI's CFO), Shivon Zilis (Musk's intermediary on the OpenAI board), and \"Self\" — Elon Musk. Bates 2024MUSK-0014185.\n\n## Logistics\n- **Trial admission:** Day 3 (April 29, 2026) — Musk direct concludes; cross begins. Centerpiece of defense's \"Musk himself was creating the for-profit\" line — used to neutralize the plaintiffs' \"stole a charity\" theory by establishing that Musk in September 2017 was personally promising the very B-corp conversion he later sued OpenAI for completing.\n- **Box upload:** 2026-04-29 15:36:42 PT — Day 3 mid-afternoon batch.\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~148 KB, 1 page.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `691 .pdf`.\n\n## Transcribed text\n\n> **Platform:** SMS-MMS-Chats\n> **Name:** 37dcb68be51233cbd95289a00f8cf798cf7e2b6aa3106f3204b2782613882394\n> **Timezone:** UTC-0700\n>\n> Participants: Sam [Altman], Jared Birchall, Shivon Zilis, Self (Elon Musk)\n>\n> ---\n>\n> **2017/09/13**\n>\n> **Elon Musk** — 07:43:00 am\n>\n> We should get going on creating the OpenAI B corp, as I promised Greg and Ilya. Let's discuss this eve. **Still no word from Sam Altman btw.**\n\n## Commentary\n\nDX 691 is, in five sentences, the spine of the defense's \"Musk wanted the for-profit\" story. **Defense's read:** in mid-September 2017 — one week before he wrote the [[PX 157]] \"final straw\" email accusing OpenAI of betraying its nonprofit mission — Musk himself was texting Birchall and Zilis to \"**get going on creating the OpenAI B corp, as I promised Greg and Ilya**.\" The text confirms (i) the existence of an affirmative promise from Musk to the founders to make the B-corp happen and (ii) Musk's personal initiative in setting it up. Plaintiffs' \"stole a charity\" theory, the defense argues, is irreconcilable with Musk's own contemporaneous willingness to sponsor the for-profit conversion. **Plaintiffs' read:** the text is a snapshot of one possible structure Musk was considering; what changed in the next seven days was the **terms of control** — the founders refused to give him 51% / CEO / unilateral authority (see [[PX 156]] / [[PX 157]] / [[Brockman Journal]]), at which point Musk pulled funding. The for-profit Musk was \"promising to create\" was one in which he would have **unequivocal control**, not the structure that ultimately emerged. The \"Still no word from Sam Altman btw\" coda is consistent with the August–September 2017 tension: Altman had stood his ground on the control question, and his silence is the silence of a partner declining to capitulate. The participant list — Birchall (CFO of the Musk Foundation), Zilis (Musk's information conduit into OpenAI), Altman — is the inner circle that runs through plaintiffs' Pioneer Building and DAF threads (see [[Key Themes]] § \"The Pioneer Building\" and § \"donor-advised funds\"). See [[Day 3|Day 3 digest]] for the cross use.\n\n---\n*See also:* [[Day 3|Day 3 digest]] · [[DX 507]] · [[DX 631]] · [[PX 156]] · [[PX 157]] · [[Jared Birchall]] · [[Shivon Zilis]] · [[Sam Altman]] · [[Brockman Journal]]\n"}
{"exhibit_id": "DX-693", "exhibit": "DX 693", "party": "Defense (OpenAI/MS)", "type": "Articles of incorporation", "admitted_trial_day": "Day 4 (April 30, 2026)", "uploaded_box_pt": "2026-04-30T15:13:13", "uploader": "Morrison Foerster", "pages": 4, "size_bytes": 2361740, "source_pdf": "DX-693.pdf", "pdf_url": "https://media.mts-in.com/DX-693.pdf", "body_markdown": "# DX 693 — Sept 15, 2017 \"Open Artificial Intelligence Technologies, Inc.\" PBC certificate (and Feb 20, 2018 dissolution)\n\n> The Delaware Public Benefit Corporation that Musk's lawyers actually filed during the September 2017 negotiation — incorporated 09/15/2017, **dissolved on the short-form \"before-beginning-business\" certificate 02/20/2018** with no capital paid and no business begun. The defense's cleanest paper-trail support for \"**for-profit entities were created on paper… no practical step was taken to convert.**\"\n\n## Document type\n**Articles of incorporation** (and short-form dissolution). Two-document compendium: (1) Delaware Certificate of Incorporation of **Open Artificial Intelligence Technologies, Inc.**, a Public Benefit Corporation, filed and delivered 8:50 PM 09/15/2017 (SR 20176192766, File No. 6544840), 12 articles, signed by incorporator Mark A. Cassanego of 216 Park Road, Burlingame, CA; and (2) Delaware Short Form Certificate of Dissolution Before Beginning Business (Sections 274 and 391(a)(5)(b)), filed 1:22 PM 02/20/2018 (SR 20181146896, same file number), signed by Cassanego as \"Majority of Incorporators or Directors.\" Bates OPENAI_MUSK00038360–00038363.\n\n## Logistics\n- **Trial admission:** Day 4 (April 30, 2026) — Musk cross + redirect; Birchall direct + cross. The for-profit-on-paper exhibit set lines up next to Birchall's Day 4 cross — see [[Key Themes#The 2018 \"purple box\" term sheet]] and the [[Key Themes]] line \"**for-profit entities were created on paper. … no practical step was taken to convert.**\"\n- **Box upload:** 2026-04-30 15:13:13 PT — Day 4 mid-afternoon batch.\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~2.25 MB, 4 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `693.pdf`.\n\n## Transcribed text\n\n> **CERTIFICATE OF INCORPORATION OF OPEN ARTIFICIAL INTELLIGENCE TECHNOLOGIES, INC. — A PUBLIC BENEFIT CORPORATION**\n> *State of Delaware, Secretary of State, Division of Corporations · Delivered 08:50 PM 09/15/2017 · FILED 08:50 PM 09/15/2017 · SR 20176192766 — File Number 6544840*\n>\n> **ARTICLE 1.** The name of this Corporation is OPEN ARTIFICIAL INTELLIGENCE TECHNOLOGIES, INC.\n>\n> **ARTICLE 2.** A. The address of this Corporation's registered office in the State of Delaware is 1209 Orange Street, in the City of Wilmington, County of New Castle, Zip Code 19801. The name of this corporation's registered agent at such address is The Corporation Trust Company. B. The name and mailing address of the incorporator of this Corporation is: Mark A. Cassanego, 216 Park Road, Burlingame, CA 94010.\n>\n> **ARTICLE 3.** The specific public benefit purpose of the corporation is to produce a public benefit and to operate in a responsible and sustainable manner that balances the stockholders' pecuniary interests, and the best interests of those materially affected by the corporation's conduct, including, but not limited to **discovering and enacting the path to safe artificial general intelligence.**\n>\n> **ARTICLE 4.** This Corporation is authorized to issue one class of stock to be designated \"Common Stock,\" with a par value of $0.001 per share. The total number of shares which the Corporation is authorized to issue is **One Million (1,000,000)**.\n>\n> **ARTICLE 5.** Except as otherwise provided in this Certificate of Incorporation, in furtherance and not in limitation of the powers conferred by statute, the board of directors is expressly authorized to make, repeal, alter, amend and rescind any or all of the Bylaws of this Corporation.\n>\n> **ARTICLE 6.** The number of directors of this Corporation shall be determined in the manner set forth in the Bylaws of this Corporation.\n>\n> **ARTICLE 7.** Elections of directors need not be by written ballot unless the Bylaws of this Corporation shall so provide.\n>\n> **ARTICLE 8.** Meetings of stockholders may be held within or without the State of Delaware, as the Bylaws may provide. The books of this Corporation may be kept (subject to any provision contained in the statutes) outside the State of Delaware at such place or places as may be designated from time to time by the board of directors or in the Bylaws of this Corporation.\n>\n> **ARTICLE 9.** A director of this Corporation shall not be personally liable to this Corporation or its stockholders for monetary damages for breach of fiduciary duty as a director, including, but not limited to, any disinterested conduct in furtherance of this Corporation's public benefit or benefits, except for liability (i) for any breach of the director's duty of loyalty to this Corporation or its stockholders, (ii) for acts or omissions not in good faith or which involve intentional misconduct or a knowing violation of law, (iii) under Section 174 of the Delaware General Corporation Law, or (iv) for any transaction from which the director derived any improper personal benefit. If the Delaware General Corporation Law is amended after approval by the stockholders of this Article to authorize corporate action further eliminating or limiting the personal liability of directors, then the liability of a director of this Corporation shall be eliminated or limited to the fullest extent permitted by the Delaware General Corporation Law as so amended. Any repeal or modification of the foregoing provisions of this Article 9 by the stockholders of this Corporation shall not adversely affect any right or protection of a director of this Corporation existing at the time of, or increase the liability of any director of this Corporation with respect to any acts or omissions of such director occurring prior to, such repeal or modification.\n>\n> **ARTICLE 10.** To the fullest extent permitted by applicable law, this Corporation is authorized to provide indemnification of (and advancement of expenses to) agents of this Corporation (and any other persons to which the Delaware General Corporation Law permits this Corporation to provide indemnification) through Bylaw provisions, agreements with such agents or other persons, vote of stockholders or disinterested directors or otherwise, in excess of the indemnification and advancement otherwise permitted by Section 145 of the Delaware General Corporation Law, subject only to limits created by applicable Delaware General Corporation Law, with respect to actions for breach of duty to this Corporation, its stockholders, and others. Any amendment, repeal or modification of any of the foregoing provisions of this Article 10 shall not adversely affect any right or protection of a director, officer, agent or other person existing at the time of, or increase the liability of any director of this Corporation with respect to any acts or omissions of such director, officer or agent occurring prior to, such amendment, repeal or modification.\n>\n> **ARTICLE 11.** This Corporation reserves the right to amend, alter, change or repeal any provision contained in this Certificate of Incorporation, in the manner now or hereafter prescribed by statute, and all rights conferred upon stockholders herein are granted subject to this reservation.\n>\n> **ARTICLE 12.** In connection with repurchases by this Corporation of its Common Stock from employees, officers, directors, advisors, consultants or other persons performing services for this Corporation or any subsidiary pursuant to agreements under which this Corporation has the option to repurchase such shares at cost upon the occurrence of certain events, such as the termination of employment, **Section 500 of the California Corporations Code shall not apply** in all or in part with respect to such repurchases.\n>\n> THE UNDERSIGNED, being the incorporator hereinbefore named, for the purpose of forming a corporation to do business both within and without the State of Delaware and in pursuance of the General Corporation Law of Delaware, does make and file this Certificate, hereby declaring and certifying that the facts herein stated are true, and accordingly has hereunto set his hand this **15th day of September, 2017.**\n>\n> /s/ MARK A. CASSANEGO, Incorporator\n> *(file ref: 40817-00005\\iManage\\9140940.1)*\n>\n> ---\n>\n> **STATE OF DELAWARE — SHORT FORM CERTIFICATE OF DISSOLUTION BEFORE BEGINNING BUSINESS (SECTIONS 274 and 391(a)(5)(b))**\n> *Delivered 01:22 PM 02/20/2018 · FILED 01:22 PM 02/20/2018 · SR 20181146896 — File Number 6544840*\n>\n> The corporation organized and existing under the General Corporation Law of the State of Delaware, hereby certifies as follows:\n>\n> 1. The name of the corporation is **OPEN ARTIFICIAL INTELLIGENCE TECHNOLOGIES, INC.**\n> 2. The date of filing of the Corporation's original Certificate of Incorporation in Delaware was **SEPTEMBER 15, 2017**.\n>\n> *(Please indicate which of the following applies by checking either 3A or 3B)*\n> **3A. [✓] No part of the capital of the Corporation has been paid.**\n> 3B. [ ] The amount of capital actually paid in for the Corporation's shares, less any part thereof disbursed for necessary expenses, has been returned to those entitled thereto.\n>\n> 4. The corporation has no assets and the business for which the corporation was organized has not begun.\n> 5. The corporation, for each year since its incorporation in this state, has been required to pay only the minimum Franchise Tax then prescribed by Section 503 of the General Corporation Law of the State of Delaware.\n> 6. The corporation has paid all fees due to or assessable by this State through the end of the year in which the certificate of dissolution is filed.\n> 7. All issued stock certificates, if any, have been surrendered or cancelled.\n> 8. All of the rights and franchises of the Corporation are hereby surrendered.\n>\n> By: /s/ MARK A. CASSANEGO, Incorporator\n> *Majority of Incorporators or Directors*\n\n## Commentary\n\nDX 693 is the most literal \"**for-profit entities were created on paper… no practical step was taken to convert**\" exhibit in the case (see [[Key Themes]]). The Sept. 15, 2017 PBC was filed five days *before* the [[PX 157]] \"final straw\" email — i.e., during the active \"Honest Thoughts\" negotiation week — with the textbook \"**discovering and enacting the path to safe artificial general intelligence**\" public-benefit purpose, and with 1,000,000 authorized common shares. The Feb. 20, 2018 short-form dissolution then certifies that **no capital was paid in**, **no business begun**, **no stock issued**. Defense uses the pair to say: when the founders couldn't agree with Musk on control, they didn't quietly convert OpenAI behind his back — they killed the empty shell and stayed in the nonprofit lane, only standing up the for-profit (OpenAI LP) months later in the [[Key Themes#The 2018 \"purple box\" term sheet|\"purple-box\"]] term-sheet round Musk now says he wasn't told about. Plaintiffs' counter on this exhibit on Day 4 cross / redirect is that filing-then-killing a Delaware PBC during the negotiation Musk was actively trying to control supports their **\"Musk wanted unequivocal control\"** read of [[PX 157]] only insofar as the founders were *prepared* to convert — but the conversion they ultimately executed in 2018-19 (per [[Key Themes#The 2018 \"purple box\" term sheet|the purple box]]) is the structure plaintiffs say breached the trust.\n\n---\n*See also:* [[Day 4|Day 4 digest]] · [[PX 157]] · [[DX-632]] · [[DX-657]] · [[DX-758]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-698", "exhibit": "DX 698", "party": "Defense", "type": "Text messages", "admitted_trial_day": "Day 4 (April 30, 2026)", "uploaded_box_pt": "2026-04-30T15:13:13", "uploader": "Morrison Foerster", "pages": 2, "size_bytes": 112902, "source_pdf": "DX-698.pdf", "pdf_url": "https://media.mts-in.com/DX-698.pdf", "body_markdown": "# DX 698 — Sept 19, 2017 Shivon Zilis ↔ Musk (\"Mr. President?\") iMessage thread: \"Seems like Greg, Ilya, Elon a go on for profit … Open Artificial Intelligence Technologies, Inc.\"\n\n> Twelve text messages on **September 19, 2017** — the day before the [[PX 157]] \"final straw\" email — between Shivon Zilis and Musk (\"Mr. President?\") in which Zilis reports that \"**Greg, Ilya, Elon a go on for profit**\" and Musk replies \"**Ok, i've already created the B corp. … Open Artificial Intelligence Technologies, Inc. … But will be referred to as Open AI. Kinda like SpaceX. … On the plane ride home with E last week. All him.**\"\n\n## Document type\n**Text messages** — Apple iMessage Short Message Report. Conversation 71a803a9f82bdf96ea70b94549ae18c5; 12 messages on 9/19/2017; participants 2 (Shivon Zilis SZ; \"Mr. President?\" P [Musk]). Bates ZILIS-0001032–0001033. Marked Confidential.\n\n## Logistics\n- **Trial admission:** Day 4 (April 30, 2026) — Musk cross + redirect; Birchall direct + cross. Used by defense to corroborate Zilis's Day 8 testimony and to ground the [[Key Themes]] §\"Unequivocal control\" theme.\n- **Box upload:** 2026-04-30 15:13:13 PT — Day 4 mid-afternoon Defense batch (clustered with [[DX 600]], [[DX 638]]).\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~110 KB, 2 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `698.pdf`.\n\n## Transcribed text\n\n> **Short Message Report**\n> Conversations: 1 | Participants: 2\n> Total Messages: 12 | Date Range: 9/19/2017\n>\n> **Outline of Conversations**\n> 71a803a9f82bdf96ea70b94549ae18c5 · 12 messages on 9/19/2017 · Undefined Participant · **Shivon Zilis**\n\n> **Messages in chronological order** (times shown in GMT +00:00)\n>\n> **9/19/2017, 6:55 PM — Shivon Zilis (SZ):** Heads up\n>\n> **9/19/2017, 6:55 PM — SZ:** **Seems like Greg, Ilya, Elon a go on for profit**\n>\n> **9/19/2017, 6:56 PM — SZ:** Sam likely an investor to start\n>\n> **9/19/2017, 6:56 PM — Mr. President? (P):** **Ok, i've already created the B corp.**\n>\n> **9/19/2017, 6:56 PM — SZ:** OpenAI?\n>\n> **9/19/2017, 6:56 PM — SZ:** Like that name?\n>\n> **9/19/2017, 6:56 PM — P:** **Open Artificial Intelligence Technologies, Inc.**\n>\n> **9/19/2017, 6:57 PM — P:** **But will be referred to as Open AI. Kinda like SpaceX.**\n>\n> **9/19/2017, 7:07 PM — SZ:** Oh I loooove that\n>\n> **9/19/2017, 7:07 PM — SZ:** You come up with that solution?\n>\n> **9/19/2017, 8:02 PM — P:** **On the plane ride home with E last week. All him.**\n>\n> **9/19/2017, 8:03 PM — P:** But I don't have any marching orders beyond that…\n\n## Commentary\n\nDX 698 is defense's cleanest exhibit for the \"**Musk wanted control / Musk was the architect of the for-profit conversion**\" theme (see [[Key Themes]] §\"Unequivocal control\"): on **September 19, 2017** — one day before he sent the [[PX 157]] \"final straw\" email walking away — Musk had already personally **incorporated** the new for-profit B Corp under the name **\"Open Artificial Intelligence Technologies, Inc.\"** and planned for it to be \"referred to as Open AI. Kinda like SpaceX.\" The \"**On the plane ride home with E last week. All him**\" line ties the naming and the conversion plan directly to Musk and to a private flight conversation with \"E\" (Ilya?). The messages corroborate the same control fight visible in [[PX 152]] (Musk's Aug 28, 2017 \"I've had enough\" reply to Zilis), [[PX 151]] (Brockman's Aug 21, 2017 \"this is the only chance we have to get out from Elon\" journal entry), and [[DX 679]] (Musk's Sept 11, 2017 \"you are pushing too hard here\" reply to Brockman's cap-table proposal). Plaintiffs' answer on cross is that incorporating a B Corp is not the same as converting OpenAI to a for-profit, and the architecture Musk described in DX 698 is the **with-Musk** structure that was on the table; the without-Musk version is what eventually happened in 2019.\n\n---\n*See also:* [[Day 4|Day 4 digest]] · [[PX 151]] · [[PX 152]] · [[PX 157]] · [[DX 679]] · [[Shivon Zilis]] · [[Elon Musk]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-704", "exhibit": "DX 704", "party": "Defense", "type": "Email thread", "admitted_trial_day": "Day 7 (May 5, 2026)", "uploaded_box_pt": "2026-05-05T14:27:00", "uploader": "Morrison Foerster", "pages": 3, "size_bytes": 508744, "source_pdf": "DX-704.pdf", "pdf_url": "https://media.mts-in.com/DX-704.pdf", "body_markdown": "# DX 704 — Sept 20, 2017 Musk → founders, \"Re: Honest Thoughts\" (defense version of the \"final straw\" thread)\n\n> The defense's version of the same Sept 20, 2017 email thread admitted as plaintiffs' [[PX 157]] — Musk's \"Guys, I've had enough. **This is the final straw**\" reply on top of Sutskever and Brockman's \"Honest Thoughts\" letter (\"**absolute control is extremely important to you**\"; \"**make ourselves easily firable — whatever it takes to work with you**\").\n\n## Document type\n**Email thread, plain text, two messages.** Same text as [[PX 157]]. From Elon Musk; to Ilya Sutskever; cc Sam Altman, Greg Brockman, Sam Teller, Shivon Zilis; subject \"Re: Honest Thoughts\"; date Wednesday, September 20, 2017 2:17:03 PM. Three pages. Bates 2024MUSK-0000224–0000226. **Note: this exhibit and PX 157 are the same email thread**, admitted by both sides under their own exhibit numbers — a common litigation pattern where each side wants its own foundation for the document.\n\n## Logistics\n- **Trial admission:** Day 7 (May 5, 2026 — Brockman defense direct + Mr. Molo recross + DX 1252; Wu deposition). Defense direct of Brockman used DX 704 to walk through the \"Honest Thoughts\" letter in Brockman's own voice as the architectural-counterweight reading: the founders' core demand on Musk was *limits on his control*, not loyalty to nonprofit form.\n- **Box upload:** 2026-05-05 14:27:00 PT — Day 7 early-afternoon defense batch (clustered with DX 1205).\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~497 KB, 3 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `0704.pdf`.\n\n## Transcribed text\n\n> **From:** Elon Musk <erm@spacex.com>\n> **To:** Ilya Sutskever <ilyasu@openai.com>\n> **Cc:** Sam Altman <shga@ycombinator.com>, Greg Brockman <gdb@openai.com>, Sam Teller <steller@spacex.com>, Shivon Zilis <shivon@openai.com>\n> **Subject:** Re: Honest Thoughts\n> **Date:** Wednesday, September 20, 2017 2:17:03 PM\n>\n> Guys, I've had enough. This is the final straw.\n>\n> Either go do something on your own or continue with OpenAI as a nonprofit. I will no longer fund OpenAI until you have made a firm commitment to stay or I'm just being a fool who is essentially providing free funding for you to create a startup.\n>\n> Discussions are over.\n\n> **From:** Ilya Sutskever <ilyasu@openai.com>\n> **To:** Elon Musk <erm@spacex.com>, Sam Altman <shga@ycombinator.com>\n> **Cc:** Greg Brockman <gdb@openai.com>, Sam Teller <steller@spacex.com>, Shivon Zilis <shivon@openai.com>\n> **Subject:** Honest Thoughts\n> **Date:** Sep 20, 2017, at 2:08 PM\n>\n> Elon, Sam,\n>\n> This process has been the highest stakes conversation that Greg and I have ever participated in, and if the project succeeds, it'll turn out to have been the highest stakes conversation the world has seen. It's also been a deeply personal conversation for all of us.\n>\n> Yesterday while we were considering making our final commitment given the non-solicit agreement, we realized we'd made a mistake. We have several important concerns that we haven't raised with either of you. We didn't raise them because we were afraid to: we were afraid of harming the relationship, having you think less of us, or losing you as partners.\n>\n> There is some chance that our concerns will prove to be unresolvable. We really hope it's not the case, but we know we will fail for sure if we don't all discuss them now. And we have hope that we can work through them and all continue working together.\n>\n> **Elon:**\n>\n> We *really* want to work with you. We believe that if we join forces, our chance of success in the mission is the greatest. Our upside is the highest. There is no doubt about that. Our desire to work with you is so great that we are happy to give up on the equity, personal control, **make ourselves easily firable** — whatever it takes to work with you.\n>\n> But we realized that we were careless in our thinking about the implications of control for the world. Because it seemed so hubristic, we have not been seriously considering the implications of success.\n>\n> - The current structure provides you with a path where you end up with **unilateral absolute control over the AGI**. You stated that you don't want to control the final AGI, but during this negotiation, you've shown to us that **absolute control is extremely important to you**.\n>   - As an example, you said that you needed to be CEO of the new company so that everyone will know that you are the one who is in charge, even though you also stated that you hate being CEO and would much rather not be CEO.\n> - Thus, we are concerned that as the company makes genuine progress towards AGI, you will choose to retain your absolute control of the company despite current intent to the contrary. We disagree with your statement that our ability to leave is our greatest power, because once the company is actually on track to AGI, the company will be much more important than any individual.\n> - The goal of OpenAI is to make the future good and to avoid an AGI dictatorship. You are concerned that Demis could create an AGI dictatorship. So do we. So it is a bad idea to create a structure where you could become a dictator if you chose to, especially given that we can create some other structure that avoids this possibility.\n>\n> We have a few smaller concerns, but we think it's useful to mention it here:\n>\n> - In the event we decide to buy Cerebras, my strong sense is that it'll be done through Tesla. But why do it this way if we could also do it from within OpenAI? Specifically, the concern is that Tesla has a duty to shareholders to maximize shareholder return, which is not aligned with OpenAI's mission. So the overall result may not end up being optimal for OpenAI.\n> - We believe that OpenAI the non-profit was successful because both you and Sam were in it. Sam acted as a genuine counterbalance to you, which has been extremely fruitful. Greg and I, at least so far, are much worse at being a counterbalance to you. We feel this is evidenced even by this negotiation, where we were ready to sweep the long-term AGI control questions under the rug while Sam stood his ground.\n>\n> **Sam:**\n>\n> When Greg and I are stuck, you've always had an answer that turned out to be deep and correct. You've been thinking about the ways forward on this problem extremely deeply and thoroughly. Greg and I understand technical execution, but we don't know how structure decisions will play out over the next month, year, or five years.\n>\n> But we haven't been able to fully trust your judgements throughout this process, because we don't understand your cost function.\n>\n> - We don't understand why the CEO title is so important to you. Your stated reasons have changed, and it's hard to really understand what's driving it.\n> - Is AGI *truly* your primary motivation? How does it connect to your political goals? How has your thought process changed over time?\n>\n> **Greg and Ilya:**\n>\n> We had a fair share of our own failings during this negotiation, and we'll list some of them here (Elon and Sam, I'm sure you'll have plenty to add…):\n>\n> - During this negotiation, we realized that we have allowed the idea of financial return 2-3 years down the line to drive our decisions. This is why we didn't push on the control — we thought that our equity is good enough, so why worry? But this attitude is wrong, just like the attitude of AI experts who don't think that AI safety is an issue because they don't really believe that they'll build AGI.\n> - We did not speak our full truth during the negotiation. We have our excuses, but it was damaging to the process, and we may lose both Sam and Elon as a result.\n>\n> There's enough baggage here that we think it's very important for us to meet and talk it out. Our collaboration will not succeed if we don't. Can all four of us meet today? If all of us say the truth, and resolve the issues, the company that we'll create will be much more likely to withstand the very strong forces it'll experience.\n>\n> - Greg & Ilya\n\n## Commentary\n\nDX 704 is the **defense's version** of the same email plaintiffs admitted as [[PX 157]] — and the duplicate admission is itself a strategic move. **Defense's read** (the reason the document is in *its* exhibit list at all): the founders' \"Honest Thoughts\" letter is plaintiffs' single best evidence *for the defense* — Sutskever and Brockman tell Musk to his face that \"**absolute control is extremely important to you**,\" that he was demanding to be CEO, and that the proposed structure would let him become \"**a dictator if you chose to**.\" Defense uses DX 704 to convert plaintiffs' \"[[Key Themes|stole a charity]]\" theme into the [[Key Themes|\"unequivocal control\"]] / \"Musk picked up his marbles\" theme — and the [[Brockman Journal|Brockman journal entries]] of the same week (PX 151 Aug 21, PX-1250-16 Aug 18) become not \"stealing the nonprofit\" entries but contemporaneous notes on how to *resist Musk's unilateral-control demand*. **Plaintiffs' read** of the same document (under PX 157): the line \"**I will no longer fund OpenAI**\" is Exhibit A for Musk treating donations as conditional, and the founders' offer to \"**give up on the equity, personal control, make ourselves easily firable**\" is the corporate-form equivalent of a fiduciary commitment they later abandoned. The Cerebras concern in the letter (\"*it'll be done through Tesla*\") is also the textual seed for plaintiffs' Cerebras-conflict-of-interest theory built around Brockman's March 2017 Cerebras stock purchase (see Day 6 testimony in [[Key Themes]]). For full duplicative analysis see [[PX 157]].\n\n---\n*See also:* [[PX 157]] · [[Day 7|Day 7 digest]] · [[PX 151]] · [[Brockman Journal]] · [[Greg Brockman]] · [[Ilya Sutskever]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-707", "exhibit": "DX 707", "party": "Defense", "type": "Email thread", "admitted_trial_day": "Day 8 (May 6, 2026)", "uploaded_box_pt": "2026-05-06T14:56:53", "uploader": "Morrison Foerster", "pages": 1, "size_bytes": 379159, "source_pdf": "DX-707.pdf", "pdf_url": "https://media.mts-in.com/DX-707.pdf", "body_markdown": "# DX 707 — Sept 26–27, 2017 Musk \"conditions of resuming funding\" — board seats demand\n\n> One-page email thread (Zilis → Musk → Teller → Musk) on Sept 26–27, 2017 in which Musk, six days after the [[PX 157]] \"final straw\" message, lays out the **price of resuming OpenAI funding**: **5 of 10 board seats** for his side, a 12-month non-solicit, and 2-year commitments from Brockman and Sutskever — \"Please let them know that this a condition of resuming funding.\"\n\n## Document type\n**Email thread, single page, four messages.** From Elon Musk <erm@spacex.com> to Sam Teller <steller@openai.com>, cc Shivon Zilis <shivon@openai.com> and Jared Birchall <jared@excession.com>, subject \"Re: OpenAI Board Update,\" dated Wed, 27 Sep 2017 01:10:01 +0000. Production-stamped Confidential; Bates 2024MUSK-0008757. Document ID DX-0707.\n\n## Logistics\n- **Trial admission:** Day 8 (May 6, 2026) — Murati live; Zilis live; Toner video begins. Per the wiki TSV: \"**DX 707 — Sept 2017 Musk conditions for resuming funding.**\" Used as part of the defense's Zilis sequence to show Musk's *terms* of resuming funding were structural-control terms, not mission-defense terms — directly continuous with the [[Key Themes#\"Unequivocal control\" (the 2017 negotiation breakdown)|\"unequivocal control\"]] line.\n- **Box upload:** 2026-05-06 14:56:53 PT — Day 8 mid-afternoon batch (clustered with DX 643, DX 728, DX 758, DX 925).\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~370 KB, 1 page.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `0707.pdf`.\n\n## Transcribed text\n\n> **From:** Elon Musk <erm@spacex.com>\n> **To:** Sam Teller <steller@openai.com>\n> **Cc:** Shivon Zilis <shivon@openai.com>, Jared Birchall <jared@excession.com>\n> **Subject:** Re: OpenAI Board Update\n> **Date:** Wed, 27 Sep 2017 01:10:01 -0000\n>\n> Yeah for now. My need to give up seats later if we get high profile AI leaders on board.\n\n> **From:** Sam Teller <steller@openai.com>\n> **To:** Elon Musk <erm@spacex.com>\n> **Cc:** Shivon Zilis <shivon@openai.com>, Jared Birchall <jared@excession.com>\n> **Subject:** Re: OpenAI Board Update\n> **Date:** Sep 26, 2017, at 5:45 PM\n>\n> To make this concrete and to give you 4/8 board seats, should we ask that Shivon, Jared, and I join the board now?\n\n> **From:** Elon Musk <erm@spacex.com>\n> **To:** Shivon Zilis <shivon@openai.com>\n> **Cc:** Sam Teller <steller@openai.com>, Jared Birchall <jared@excession.com>\n> **Subject:** Re: OpenAI Board Update\n> **Date:** Tue, Sep 26, 2017 at 5:16 PM\n>\n> Board seats need to at least roughly match the proportion of funding provided, rounding down for bigger funders and up for smaller. We should have at least 5/10 seats, which is still way less than my funding percentage at 2/3. I don't need a majority, but will be ok with half the board vote.\n>\n> Please let them know that this a condition of resuming funding. That and a reasonable non-solicit for 12 months. Greg and Ilya would have to commit to at least two years.\n\n> **From:** Shivon Zilis <shivon@openai.com>\n> **To:** Elon Musk <erm@spacex.com>\n> **Cc:** Sam Teller <steller@openai.com>, Jared Birchall <jared@excession.com>\n> **Subject:** OpenAI Board Update\n> **Date:** Sep 26, 2017, at 4:56 PM\n>\n> Chris Clark has resigned from the OpenAI board. By Greg's, Sam's, and his own account, he was only ever on for administrative purposes at inception, so the removal was always intended (and overdue) but now official.\n>\n> The board is now:\n>\n> Elon Musk\n> Sam Altman\n> Greg Brockman\n> Ilya Sutskever\n> Holden Karnofsky (Open Philanthropy)\n\n## Commentary\n\nDX 707 sits between [[PX 157]] (Sept 20, \"final straw\") and the eventual collapse of the for-profit conversion negotiation a few weeks later. It is the **price list** Musk's side put on the table: half the board, plus binding multi-year commitments from Brockman and Sutskever, plus a non-solicit. The \"**condition of resuming funding**\" phrase is the defense's strongest direct quotation for the [[Key Themes#\"Unequivocal control\" (the 2017 negotiation breakdown)|\"unequivocal control\"]] frame — Musk himself, in writing, treating his donations as the lever for governance terms (\"**at least 5/10 seats**\" / \"**half the board vote**\") and personnel commitments. The exhibit also establishes Musk's *implicit* admission against interest on the Brockman-zero-contribution thread: he himself says board seats should map to **funding**, not mission contribution — which is precisely the metric Birchall later confirmed at trial gives Brockman a board seat for **zero** dollars contributed (`043026TT.txt:4416`; see [[Key Themes#Greg Brockman's contributions (\"zero\")]]). On plaintiffs' read, the exhibit is consistent with Musk's \"I was a fool\" theme: he was extracting commitments to *protect* the nonprofit. On defense's read, it pairs with [[PX 157]]'s recipients-side response — \"you've shown to us that absolute control is extremely important to you\" — to make the \"Musk wanted to control AGI himself\" story.\n\n---\n*See also:* [[Day 8|Day 8 digest]] · [[Shivon Zilis]] · [[PX 157]] · [[PX 156]] · [[DX 643]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-708", "exhibit": "DX 708", "party": "Defense", "type": "Email", "admitted_trial_day": "Day 8 (May 6, 2026)", "uploaded_box_pt": "2026-05-06T14:56:53", "uploader": "Morrison Foerster", "pages": 1, "size_bytes": 413965, "source_pdf": "DX-708.pdf", "pdf_url": "https://media.mts-in.com/DX-708.pdf", "body_markdown": "# DX-708 — Sept 28, 2017 Teller-to-Zilis email summarizing Musk's three demands to Greg & Ilya\n\n> Sam Teller's debrief of his Sept. 28, 2017 conversation with Brockman and Sutskever — listing Musk's pre-conditions: control via \"equal seats,\" a two-year commitment, and a non-solicit.\n\n## Document type\n**Email, plain text.** Single message from Sam Teller (steller@openai.com) to Shivon Zilis (shivon@tesla.com), Re: \"Greg and Ilya,\" 9/28/2017 1:37:03 AM. Replies on top of an earlier 6:05 PM email from Zilis. Bates TESLA_000002281. Marked HIGHLY CONFIDENTIAL — AEO.\n\n## Logistics\n- **Trial admission:** Day 8 (May 6, 2026) — Murati live; Zilis live; Toner video begins.\n- **Box upload:** 2026-05-06 14:56:53 PT — Day-8 batch.\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~414 KB, 1 page.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `0708.pdf`. Wiki reference: \"DX 708 — Sept 2017 conditions.\"\n\n## Transcribed text\n\n> **From:** Sam Teller <steller@openai.com>\n> **To:** Shivon Zilis <shivon@tesla.com>\n> **Subject:** Re: Greg and Ilya\n> **Date:** 9/28/2017 1:37:03 AM\n>\n> Cool\n>\n> Thank you\n\n> **From:** Shivon Zilis <shivon@tesla.com>\n> **To:** Sam Teller <steller@openai.com>\n> **Subject:** Greg and Ilya\n> **Date:** Sep 28, 2017, at 6:05 PM\n>\n> (nothing big here and I know we are tabling until next week or at least after IAC, but if you want anything shared lmk)\n>\n> Shared Elon's rough expectations for them to think about in advance of next week. To be precise: I did not say that you / me / Jared may take seats but I shared that he would need 1) **control in the form of equal seats for Elon vs. rest of seats** 2) **two year commitment** 3) **a reasonable non-solicit**.\n>\n> They took it in stride… I just said I thought they should internalize it and at least think through all the best possible ways this could play out practically vs. getting stuck in philosophical platitude land like the last two months.\n>\n> For your knowledge:\n>\n> **Control:**\n> They claim that the *only* non-starter for them in the for-profit or now in the non-profit the transfer to a long-term control structure that included many people, not just Elon. They were scared it wouldn't happen and needed an \"iron clad\" structure that couldn't be gamed. I said it was on them to figure out what would satisfy this requirement for them (and also gave personal advice to stop coming up with general complaints about things without having any idea of what solution they would want to solve the problem).\n>\n> **Structure:**\n> They were like, \"well it sounds like he expects that all of the things he wanted for the for-profit he also wants for the non-profit, which sucks because the for profit was clearly the better structure. If that's the case is he still open to the for-profit?\" I said I had no idea but said I personally thought it was on them to think through whether there was concrete and simple solution they could propose on the long-term control front that satisfies all their anxieties before even raising such a thing, since moot point otherwise.\n>\n> **No-Go:**\n> On the above, they said there is some probability that their long-term control requirement may just be fundamentally incompatible with Elon's desire for short term control because you can't guarantee transfer. In that case they said they wouldn't be able to move forward in any structure. I said that seemed silly to me since I see no reason why a clause for transfer of majority control after 2-3 years that was well thought through wouldn't work. They said they'd have to think about it and perhaps get Sam A's thoughts on it.\n>\n> **Symmetry:**\n> Said if they went all and committed to not doing anything else for two years they'd want to know that Elon wouldn't spin up another AGI thing for next two years too. I said I obviously couldn't speak on behalf of him but that if they decided they are all in they could raise it.\n>\n> **Sam A:**\n> They said they think Sam is hopeful Elon will still want to work with OpenAI and actually really likes the non-profit structure for now. I asked on their guess on how likely the giant Azure deal was and they guessed 25% but said we probably wouldn't know the outcome for a couple of months.\n\n## Commentary\n\nDX 708 is the rosetta stone for the September 2017 control negotiation: it converts the abstract \"absolute control\" complaint of [[PX 157]] into a concrete three-bullet list — equal board seats, two-year commitment, non-solicit — relayed through Musk's chief of staff. It fills in what Sutskever and Brockman were responding to in \"Honest Thoughts,\" and it captures a foreshadowing inversion of Musk's later \"non-solicit\" complaint: here Musk is the one demanding the non-solicit (later weaponized in [[Day 3|Day 3]] cross via DX 761, \"actively try to move three or four people from OpenAI to Tesla\"). The defense uses it on [[Day 8|Day 8]] (Zilis live) for the [[Key Themes]] \"unequivocal control\" arc and to show the founders' fear of an Elon-controlled \"iron clad\" structure that \"couldn't be gamed\" was articulated in real time, not retconned.\n\n---\n\n*See also:* [[Day 8|Day 8 digest]] · [[PX 157]] · [[PX 151]] · [[Shivon Zilis]] · [[Greg Brockman]] · [[Ilya Sutskever]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-710", "exhibit": "DX 710", "party": "Defense", "type": "Email", "admitted_trial_day": "Day 8 (May 6, 2026)", "uploaded_box_pt": "2026-05-06T14:56:53", "uploader": "Morrison Foerster", "pages": 1, "size_bytes": 365307, "source_pdf": "DX-710.pdf", "pdf_url": "https://media.mts-in.com/DX-710.pdf", "body_markdown": "# DX 710 — Sept 28, 2017 Zilis → Birchall, OpenAI control conditions\n\n> A single-page Zilis email to Birchall summarizing what the OpenAI founders were demanding in late September 2017 — \"**they will not move forward without a guarantee to switch away from him having control**\" — including a 2-year majority-control sunset and the founders' \"scared that even *they* may not want to give up majority control over something as powerful as potential AGI\" rationale.\n\n## Document type\n**Email, plain text, single message.** Zilis to Birchall (jared@excession.com), September 28, 2017 9:52:18 PM. Subject \"OpenAI.\" Bates OPENAI_MUSK00023783. Document ID DX-0710. Admitted with **limiting instruction** (per [[Key Themes]]: \"DX 710 — Sept 2017 conditions (limiting instruction)\") — likely admitted for a non-truth purpose (e.g., effect on Musk-side listener / state of mind), not for the truth of the founders' stated reasons.\n\n## Logistics\n- **Trial admission:** Day 8 (May 6, 2026) — used during Zilis's live testimony. Admitted with a **limiting instruction**.\n- **Box upload:** 2026-05-06 14:56:53 PT — Day 8 mid-afternoon batch (clustered with DX 766).\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~357 KB, 1 page.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `0710.pdf`.\n\n## Transcribed text\n\n> **From:** Shivon Zilis <shivon@openai.com>\n> **To:** Jared Birchall <jared@excession.com>\n> **Subject:** OpenAI\n> **Date:** 9/28/2017 9:52:18 PM\n>\n> Hey! No need to trouble you to call, know you're super busy. Here is what you need to know:\n>\n> They say they will not move forward without a guarantee to switch away from him having control. You and I can argue that's stupid all we want, but they are holding firm on it. Potential paths forward:\n>\n> 1) A clause that guarantees after 2 years no majority control, regardless of funding ratios\n> 2) E doesn't contribute right now and it's up to them to find funding before setting the board so he's never in majority control\n> 3) They agree to disagree and we problem solve around how to make the breakup as amicable as possible\n> 4) Jared's creative ideas\n>\n> The \"trusting\" him without legal language thing is just not currently an option for them because they a) don't have the level of extreme trust you or I have for E and b) they are, in their words, scared that even *they* may not want to give up majority control over something as powerful as potential AGI if it's developed so they would want a structure that guaranteed distribution of power no matter who is in control.\n>\n> ~s\n\n## Commentary\n\nDX 710 is one of the most quotable defense exhibits in the chunk. Eight days after [[PX 157]]'s \"**final straw**\" email, Zilis — Musk's hand-picked liaison — writes Birchall confirming that the founders **\"will not move forward without a guarantee to switch away from him having control,\"** offering a menu of fall-back structures (a 2-year majority-control sunset; Musk simply not contributing while the board is set), and recording that \"**the 'trusting' him without legal language thing is just not currently an option for them.**\" Defense uses this email both to corroborate Sutskever and Brockman's [[PX 157]] account that Musk's demand for unilateral control was the deal-breaker, and to put the plaintiffs in the unusual position of impeaching their *own* Musk-side messenger — Zilis, then on the OpenAI payroll but reporting to Birchall. The **limiting instruction** tag in [[Key Themes]] is important: the court likely admitted DX 710 not for the truth of what \"they\" thought (hearsay) but for its effect on Musk and Birchall as listeners, or as evidence of the state of negotiations. The \"scared that even *they* may not want to give up majority control over something as powerful as potential AGI\" rationale matches the AGI-dictatorship framing in [[PX 157]] (\"**you could become a dictator if you chose to**\"). Cross-reference: [[DX 646]] (the Aug 16 \"OpenAI conversion\" memo) and [[DX 679]] (the Sept 11 \"you are pushing too hard\" rejection) bracket this email on either side.\n\n---\n*See also:* [[Day 8|Day 8 digest]] · [[Shivon Zilis]] · [[Jared Birchall]] · [[PX 157]] · [[DX 646]] · [[DX 679]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-712", "exhibit": "DX 712", "party": "Defense", "type": "Email", "admitted_trial_day": "Day 8 (May 6, 2026)", "uploaded_box_pt": "2026-05-06T14:56:53", "uploader": "Morrison Foerster", "pages": 2, "size_bytes": 449786, "source_pdf": "DX-712.pdf", "pdf_url": "https://media.mts-in.com/DX-712.pdf", "body_markdown": "# DX 712 — Oct 1–2, 2017 Zilis → Teller \"Re: OpenAI notes\" (long-term control / B-Corp / non-profit \"experiment\")\n\n> Shivon Zilis's October 1, 2017 8:10 PM \"honest reflection\" memo to Sam Teller — and Teller's October 2 4:10 AM reply (\"**I think better to talk through this one live**\") — laying out, post-[[PX 157]] \"final straw,\" the founders' three ranked structural options: (1) revive the B-Corp idea; (2) non-profit fundraising \"experiment\"; or, as Sam Altman's preferred path, a one-month pause on actions and funding before resuming the non-profit.\n\n## Document type\n**Email thread, plain text, two messages.** Sam Teller's October 2, 2017 3:23:09 AM reply on top of Zilis's October 1, 2017 8:10 PM original. Subject \"Re: OpenAI notes (Question for Sam on top!).\" Bates OPENAI_MUSK00003882–3883. Document ID DX-0712.\n\n## Logistics\n- **Trial admission:** Day 8 (May 6, 2026) — used during Zilis live direct. No prior wiki reference; admission day inferred from Box upload date.\n- **Box upload:** 2026-05-06 14:56:53 PT — Day 8 mid-afternoon defense batch (clustered with [[DX 627]], [[DX 651]], [[DX 1017]], [[DX 1290]]).\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~439 KB, 2 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `0712.pdf`.\n\n## Transcribed text\n\n> **From:** Sam Teller <steller@openai.com>\n> **To:** Shivon Zilis <shivon@openai.com>\n> **Subject:** Re: OpenAI notes (Question for Sam on top!)\n> **Date:** 10/2/2017 3:23:09 AM\n>\n> I think better to talk through this one live\n\n> **From:** Shivon Zilis <shivon@openai.com>\n> **To:** Sam Teller <steller@openai.com>\n> **Subject:** OpenAI notes (Question for Sam on top!)\n> **Date:** Oct 1, 2017, at 8:10 PM\n>\n> (below is an honest reflection of their views. i am a little worried to share because i don't want to seem like i'm pushing for the b corp or blow this up in their faces but lmk if ok to send. in all honesty, i think literally all of this is putting the cart before the horse because Elon doesn't even yet know if he thinks accelerating AGI is a good thing or not, but it's not my job to say things like that)\n>\n> **Expectations**\n>\n> -Greg and Ilya are aware of the expectations for you to continue funding the non-profit (equal number of board seats / votes for you vs. rest of board, a two year commitment from them, and a reasonable non-solicit).\n>\n> **Long term control**\n>\n> -Greg, Ilya, and Sam all claim the only thing they really care about is there being a long-term control provision that transitions away from short-term majority control at some defined point and ensures a broad group governance of AGI if it's created. This applied not only to the for-profit negotiation but also to the non-profit going forward. It seems like a simple clause could satisfy this if everyone agrees this is ok?\n>\n> **Greg and Ilya's ideal outcomes (in order)**\n>\n> *1. Revive the B-Corp idea*\n> -Rationale: Better from an IP / recruiting / fundraising perspective. Greg feels regret because he could \"smell blood\" re: DeepMind. \"We are now legit enough, we should have had the money, and we would have had the equity to get the right people. We even had the right list there and ready\".\n> Would require a clause that ensures that, without doubt, control transitions to a broader board governance structure after two years. (Note: they are not assuming you are open to this idea, but they did share that they regret some of their decisions and think it's the better structure)\n>\n> *2. Non-profit fundraising \"experiment\"*\n> -Rationale: They think non-profit is a viable structure but only works to serve the mission if it can get 100's of millions in donations, since compute needs and salaries in absence of equity will be costly. Therefore they'd be curious to spend a couple of months trying to see if it's possible to fundraise a sufficient amount before everyone commits to the structure. They know that you would not fund again until your conditions met.\n>\n> **Sam's ideal outcomes (in order)**\n>\n> *1. Non-profit, one month pause on any actions and funding.*\n> -Rationale: He likes the non-profit structure more than Greg and Ilya. Team took a morale hit when for-profit didn't happen / all the uncertainty around the process. He has a preference to get the team back in order vs. debating structure (just found out he has done one on ones with whole team this week). He wants Greg and Ilya to focus on leading the team to a couple of short-term wins on the research front since they were distracted for so long. He knows there will not be any funding from you during that time.\n>\n> *2. Non-profit, agree to terms and resume funding now.*\n> -He says he has no issues with the board parity requirement. If Greg and Ilya are willing to commit to the other requirements, he is a go.\n>\n> *3. B-Corp.*\n> -Is game to do it full time in President / COO role so long as there is a long term control provision. Same requirements as above.\n\n## Commentary\n\nDX 712 is the **post-[[PX 157|\"final straw\"]] cleanup memo** — written eleven days after Musk's \"Discussions are over\" ultimatum and Altman's \"i remain enthusiastic about the non-profit structure!\" reply ([[PX 158]]). It is plaintiffs' and defense's exhibit at once: **plaintiffs** read it as showing Altman *did* prefer the non-profit structure (\"he likes the non-profit structure more than Greg and Ilya\"), and that Brockman/Sutskever — not Altman — were the for-profit pushers; **defense** uses it for two more aggressive points: (i) it confirms the founders' top-ranked outcome was *still* a **B-Corp** with a long-term control transition — i.e., the eventual 2018 LP / 2024 benefit-corp arc was openly previewed to Musk's circle in October 2017, *before* Musk's [[PX 24]] Charter ratification and *years* before the LP launch; and (ii) Zilis's parenthetical — **\"Elon doesn't even yet know if he thinks accelerating AGI is a good thing or not\"** — is defense's punchline against Musk's later \"phase three\" disillusionment narrative: Musk's own Family-Office staffer, in 2017, was unsure he had a settled view on AGI direction at all. Cross-reference: [[DX 627]] and [[DX 651]] for the antecedent Zilis briefings; [[PX 359]] for the 2023 Charter 2.0 / benefit-corp draft that revives this exact framework six years later.\n\n---\n*See also:* [[Day 8|Day 8 digest]] · [[Shivon Zilis]] · [[Sam Teller]] · [[Sam Altman]] · [[PX 157]] · [[PX 158]] · [[PX 359]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-715", "exhibit": "DX 715", "party": "Defendants", "type": "Email thread", "admitted_trial_day": "Day 8 (May 6, 2026)", "uploaded_box_pt": "2026-05-06T14:56:53", "uploader": "Morrison Foerster", "pages": 1, "size_bytes": 390050, "source_pdf": "DX-715.pdf", "pdf_url": "https://media.mts-in.com/DX-715.pdf", "body_markdown": "# DX 715 — Oct 20, 2017 Sam Teller-Shivon Zilis email \"Re: Ilya\": \"Tesla solves the funding issue immediately. … Tesla at least has option to bury.\"\n\n> Limiting-instruction exhibit. Internal email between Sam Teller (Musk's Chief of Staff at the time) and Shivon Zilis (Musk's intermediary on the OpenAI board) discussing Ilya Sutskever's preference for keeping OpenAI a nonprofit — and Zilis's reply that \"Tesla solves the funding issue immediately. … Tesla at least has option to bury.\"\n\n## Document type\n**Email thread, plain text, two messages.** Sam Teller's Oct 20, 2017 1:45 PM PDT reply on top of Shivon Zilis's Oct 20, 2017 12:24 AM email. Subject: \"Re: Ilya.\" From: Sam Teller <steller@openai.com> to Shivon Zilis <shivon@openai.com>. Bates OPENAI_MUSK00023876.\n\n## Logistics\n- **Trial admission:** Day 8 (May 6, 2026) — Murati live; Zilis live; Toner video begins. Per the chunk wiki note: \"**DX 715 — limiting-instruction exhibit**.\" A core piece of the defense's \"parallel AGI play\" narrative — Zilis's own contemporaneous writing that **Tesla** would solve OpenAI's funding problem and could \"bury\" the work for stealth advantage. This is the same theme captured in [[Key Themes]] § \"'Bury this in Tesla for stealth advantage'\" and at DX 758. The court admitted it with a limiting instruction concerning Zilis's role and her later status as a former plaintiff in this case.\n- **Box upload:** 2026-05-06 14:56:53 PT — Day 8 mid-afternoon batch.\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~381 KB, 1 page.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `0715.pdf`.\n\n## Transcribed text\n\n> **From:** Sam Teller <steller@openai.com>\n> **To:** Shivon Zilis <shivon@openai.com>\n> **Subject:** Re: Ilya\n> **Date:** 10/20/2017 1:45:33 PM\n>\n> Interesting. We need to press him further. I will try to get clarity from Elon and will call you on Recruiting/PR strategy within Tesla and long term control (turned over to the people ever?)\n>\n> **We need to try a little more to convince Ilya and Greg of this path but then we will be more firm and tell them it's happening regardless.**\n\n> **From:** Shivon Zilis <shivon@openai.com>\n> **To:** Sam Teller <steller@openai.com>\n> **Subject:** Ilya\n> **Date:** Fri, Oct 20, 2017 at 12:24 AM\n>\n> He asked to talk about Rex but just ended up catching up with him informally. His thinking out loud below:\n>\n> -**His first preference is to stick with non-profit, if able to raise sufficient funds.** Rationale is easiest to keep clean moral high ground for recruiting in the non-profit structure. Is still struggling to understand the recruiting story at Tesla.\n>\n> -I asked how, in that scenario, he would figure out if they could raise enough funds? He said give it best go until 2018 with Gates types and if not do something else. He acknowledges that he doesn't have good instincts here and it's somewhat likely to fail. **I shared that my personal concern was I think a one year experiment that will probabilistically fail seems too costly given we only have 3-5 years! Tesla solves the funding issue immediately.**\n>\n> -He thought as people get more scared they will find OpenAI more and timing right for that. My personal take was there is a very small window for that because people generally need to be very scared to take real action, and **by the time people are very scared it will probably become nationalized if it's in the open. Tesla at least has option to bury.**\n\n## Commentary\n\nDX 715 is the cleanest piece of trial evidence that the **Tesla-absorbs-OpenAI scenario was being actively pushed by Musk's deputies in late 2017**, against the express resistance of Ilya Sutskever, who told Zilis that \"**his first preference is to stick with non-profit, if able to raise sufficient funds**\" because it gave OpenAI the \"**clean moral high ground for recruiting**.\" Zilis's own private response — \"Tesla solves the funding issue immediately\" and \"Tesla at least has option to bury\" — is exactly the \"**bury this in Tesla for stealth advantage**\" frame she would use to Musk in February 2018 (see [[Key Themes]] § \"'Bury this in Tesla for stealth advantage'\"). **Defense's read:** the email matters because Zilis is a *defense* witness on Day 8 (and a former plaintiff in this very case) whose contemporaneous writing positions the for-profit conversion question as a **funding** question with two paths (raise nonprofit money from Gates-types or pull OpenAI under Tesla), not as a betrayal of charitable mission. **Plaintiffs' read:** Sutskever's \"first preference\" was the nonprofit, full stop; what Teller and Zilis are conspiring to do — \"**we need to try a little more to convince Ilya and Greg of this path but then we will be more firm and tell them it's happening regardless**\" — is the *opposite* of a fiduciary acting in good faith. Tied to [[DX 758]] (the OpenAI possible scenarios deck Schubert later put to Zilis on Day 8) and to Karpathy's \"OpenAI to attach to Tesla as its cash cow\" line in [[Key Themes]] § \"Tesla / xAI as Musk's own AGI play.\" The court's limiting instruction means the jury was told to consider Zilis's email as evidence of Zilis's contemporaneous belief and not as proof of any fact about Musk's intent.\n\n---\n*See also:* [[Day 8|Day 8 digest]] · [[DX 758]] · [[Shivon Zilis]] · [[Sam Teller]] · [[Ilya Sutskever]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-716", "exhibit": "DX 716", "party": "Defense (OpenAI/MS)", "type": "Email thread", "admitted_trial_day": "Day 8 (May 6, 2026)", "uploaded_box_pt": "2026-05-06T14:56:53", "uploader": "Morrison Foerster", "pages": 2, "size_bytes": 425402, "source_pdf": "DX-716.pdf", "pdf_url": "https://media.mts-in.com/DX-716.pdf", "body_markdown": "# DX 716 — Oct 21–22, 2017 Zilis / Teller \"Greg + Ilya\" thread\n\n> Shivon Zilis and Sam Teller's October 2017 back-and-forth about whether to pull the OpenAI work into Tesla \"with or without them,\" and Zilis's enumerated rationale for why Brockman and Sutskever \"clearly prefer the non-profit, if possible.\"\n\n## Document type\n**Email thread, plain text, four messages spanning Oct 21–22, 2017** between Shivon Zilis (shivon@openai.com) and Sam Teller (steller@openai.com), subject \"Re: Greg + Ilya.\" Bates OPENAI_MUSK00023889–23890. Production-stamped CONFIDENTIAL.\n\n## Logistics\n- **Trial admission:** Day 8 (May 6, 2026) — Murati live; Zilis live; Toner video begins. Used by defense (Schubert) on Zilis cross alongside the [[DX-758|Feb 13, 2018 \"Possible Scenarios\" memo]] to develop the [[Key Themes#Tesla / xAI as Musk's own AGI play (the parallel)|\"parallel AGI play\"]] / \"burying this in Tesla for stealth advantage\" theme.\n- **Box upload:** 2026-05-06 14:56:53 PT — Day 8 mid-afternoon batch (clustered with DX 754, DX 758, DX 835, DX 877).\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~415 KB, 2 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `0716.pdf`.\n\n## Transcribed text\n\n> **From:** Sam Teller <steller@openai.com>\n> **To:** Shivon Zilis <shivon@openai.com>\n> **Subject:** Re: Greg + Ilya\n> **Date:** 10/22/2017 5:15:54 PM\n>\n> Yes\n\n> **From:** Shivon Zilis <shivon@openai.com>\n> **To:** Sam Teller <steller@openai.com>\n> **Subject:** Re: Greg + Ilya\n> **Date:** Oct 22, 2017, at 10:15 AM\n>\n> Is it useful or not useful to start thinking about how this would work at Tesla, with or without them? Their default will be to put this off as long as possible so if I can help with generating certain types of momentum lmk!\n\n> **From:** Sam Teller <steller@openai.com>\n> **To:** Shivon Zilis <shivon@openai.com>\n> **Subject:** Re: Greg + Ilya\n> **Date:** Sat, Oct 21, 2017 at 4:27 PM\n>\n> Then they can stay at the non profit…\n\n> **From:** Shivon Zilis <shivon@openai.com>\n> **To:** Sam Teller <steller@openai.com>\n> **Subject:** Re: Greg + Ilya\n> **Date:** Oct 21, 2017, at 2:42 PM\n>\n> Yes. I mentioned that Elon seems increasingly drawn to the Tesla structure as the only logical answer to creating a true Google counterpoint and I completely understand why.\n>\n> They are really scared of losing their baby and definitely open to but not warm on the Tesla thing.\n>\n> I am worried that the Rex thing shifts momentum in the wrong direction for the three of them.\n\n> **From:** Sam Teller <steller@openai.com>\n> **To:** Shivon Zilis <shivon@openai.com>\n> **Subject:** Re: Greg + Ilya\n> **Date:** Oct 21, 2017, at 2:11 PM\n>\n> Have you hinted that it will happen regardless of their support for it?\n\n> **From:** Shivon Zilis <shivon@openai.com>\n> **To:** Sam Teller <steller@openai.com>\n> **Subject:** Greg + Ilya\n> **Date:** Oct 21, 2017, at 1:09 PM\n>\n> So clearly prefer the non-profit, if possible.\n>\n> Rationale:\n>\n> 1) Recruiting story + moral high ground\n> -They think it's easy to explain why someone would choose OpenAI over Google or Deepmind. They think non-profit for good an easy story to sell but harder at a corporation. Why Tesla AI vs Google AI?\n> (I shared my thoughts on this, which is Tesla is probably the most for the good of humanity corporation that exists, and we're lucky for that, but not sure how much that swayed them)\n>\n> 2) They think it's possible to raise 100's of millions in donations.\n> -They they should try that first. Don't think fundraising has been done in earnest. They asked my opinion. I said if 20-50M a year was the right number probably all good, but I thought it was improbable to raise 100M-500M per year, and that may not be enough to do this right. They are a little naive on this one… haven't ever tried to fundraise.\n>\n> 3) They haven't internalized the advantages of burying this in Tesla for stealth advantage.\n> -They are not naturally hard wired as maneuverers. They have trouble conceiving why this is an advantage… \"what matters is that we build something meaningful\". My take was it doesn't matter if not a legit counterpoint to Google but I don't think they really see the world that way. They have trouble thinking about relative positioning.\n\n## Commentary\n\nDX 716 is the October 2017 antecedent to [[DX-758]]: months before Zilis enumerated the nine \"Possible Scenarios\" for Musk in February 2018, she was already telling Teller that Brockman and Sutskever \"clearly prefer the non-profit, if possible\" and that Musk \"seems increasingly drawn to the Tesla structure as the only logical answer to creating a true Google counterpoint.\" The phrase \"**burying this in Tesla for stealth advantage**\" — which Zilis would re-use in her Feb 2018 message to Musk (see [[Key Themes#\"Bury this in Tesla for stealth advantage\"]]) — first appears here as her own characterization of an advantage the OpenAI founders did not \"internalize.\" Defense uses the thread to show Tesla-AGI thinking was already operational in Musk's orbit weeks after the [[PX-157|Sept 20, 2017 \"final straw\"]] blowup and before any \"stole a charity\" breach plaintiffs allege. Plaintiffs' read is the inverse: Brockman and Sutskever were still telling Musk's emissaries they wanted the nonprofit and a fundraising-first plan — the same \"morally bankrupt to convert to B Corp without him\" instinct that surfaces three weeks later in the [[Brockman Journal]]. See [[Day 8|Day 8 digest]] for Zilis's live cross.\n\n---\n*See also:* [[Day 8|Day 8 digest]] · [[Shivon Zilis]] · [[DX-758]] · [[DX 761]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-719", "exhibit": "DX 719", "party": "Defense", "type": "Email", "admitted_trial_day": "Day 3 (April 29, 2026)", "uploaded_box_pt": "2026-04-29T16:15:10", "uploader": "Morrison Foerster", "pages": 1, "size_bytes": 360505, "source_pdf": "DX-719.pdf", "pdf_url": "https://media.mts-in.com/DX-719.pdf", "body_markdown": "# DX 719 — Oct 23, 2017 Musk → Ben Rapoport (Neuralink): \"Hire independently or directly from OpenAI. I have no problem if you pitch people at Open AI to work at Neuralink.\"\n\n> The one-line forwarded email — **eight days before Musk's December 2017 OpenAI board departure** — in which Musk authorizes Ben Rapoport at Neuralink to **directly poach OpenAI engineers**.\n\n## Document type\n**Email** (forwarded message), plain text. Original from Elon Musk <erm@neuralink.com> → Ben Rapoport <ben@neuralink.com>, Mon, Oct 23, 2017 at 7:40 PM, Subject \"Re: Minimally Invasive Neural Arrays Update.\" Bates Neuralink_0000549. Marked HIGHLY CONFIDENTIAL — AEO.\n\n## Logistics\n- **Trial admission:** Day 3 (April 29, 2026) — Musk direct concludes; cross begins. Per the wiki reference index: \"DX 719 — Oct 2017 Neuralink-poaching authorization.\" Used by defense in the [[Key Themes]] §\"Tesla / xAI as Musk's own AGI play\" / \"parallel AGI play\" line on cross to show that Musk himself was actively converting OpenAI into a talent pipeline for his other ventures while he was still on OpenAI's board.\n- **Box upload:** 2026-04-29 16:15:10 PT — Day 3 late-afternoon Defense batch.\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~352 KB, 1 page.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `719.pdf`.\n\n## Transcribed text\n\n> ---------- Forwarded message ---------\n> **From:** Elon Musk <erm@neuralink.com>\n> **To:** Ben Rapoport <ben@neuralink.com>\n> **Subject:** Re: Minimally Invasive Neural Arrays Update\n> **Date:** Mon, Oct 23, 2017 at 7:40 PM\n>\n> **Hire independently or directly from OpenAI. I have no problem if you pitch people at Open AI to work at Neuralink.**\n\n## Commentary\n\nDX 719 is the cleanest single document defense has for the [[Key Themes]] §\"Tesla / xAI as Musk's own AGI play\" theme: in October 2017, while Musk was **still on OpenAI's board of directors** and while Brockman was writing the August–November 2017 journal entries (see [[Brockman Journal]]) about whether Musk's control demands made the founders' position untenable, Musk was simultaneously **green-lighting Neuralink poaching of OpenAI engineers** — using his @neuralink.com address, no less. The exhibit prefigures Musk's February 2018 \"**actively try to move three or four people from OpenAI to Tesla**\" instruction to Zilis (DX 761) and the \"**poach list**\" Karpathy compiled while Musk was still on the OpenAI board (testified to by Zilis on Day 8 — see [[Key Themes]] §\"Bury this in Tesla for stealth advantage\"). Plaintiffs' response on cross is that authorizing Neuralink hiring is not the same as authorizing the looting of the nonprofit; defense's response is that the conduct establishes an irreducible inconsistency in plaintiffs' \"Musk was the only loyal nonprofit defender\" narrative.\n\n---\n*See also:* [[Day 3|Day 3 digest]] · [[Brockman Journal]] · [[Shivon Zilis]] · [[Elon Musk]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-724", "exhibit": "DX 724", "party": "Defense", "type": "Email thread", "admitted_trial_day": "Day 8 (May 6, 2026)", "uploaded_box_pt": "2026-05-06T14:56:56", "uploader": "Morrison Foerster", "pages": 4, "size_bytes": 3130216, "source_pdf": "DX-724.pdf", "pdf_url": "https://media.mts-in.com/DX-724.pdf", "body_markdown": "# DX 724 — Nov 27, 2017 Zilis ↔ O'Brien, \"Re: Tesla AI Event (First Draft of FAQs)\"\n\n> A four-page email thread from November 27, 2017 — Shivon Zilis circulating \"a first stab at an FAQ for thinking through the Tesla AI event,\" including the line \"**Elon's name is also associated with artificial intelligence more broadly, but that is primarily through OpenAI** … those who want to work on large scale AI research don't currently think of Tesla, and Elon wants to change that by announcing his intention to create a world class AI lab.\"\n\n## Document type\n**Email thread + draft FAQ document.** Sent from Shivon Zilis [tesla.com] at 11/27/2017 3:02:23 PM, to Sarah O'Brien [sobrien@tesla.com], subject \"Re: Tesla AI Event (First Draft of FAQs).\" Cc on the underlying chain: Sam Teller, Keely Sulprizio, Jenna Lutwin (all tesla.com). Marked HIGHLY CONFIDENTIAL – AEO. Bates TESLA_000002597–2600.\n\n## Logistics\n- **Trial admission:** Day 8 (May 6, 2026 — Murati live; Zilis live; Toner video begins). No prior wiki citation; admission day inferred from Box upload and Zilis's live testimony. Used by defense in connection with the [[Key Themes|\"Tesla / xAI as Musk's own AGI play\"]] thread — the FAQ explicitly frames the planned Tesla AI event as Tesla announcing \"**his intention to create a world class AI lab**\" to compete with Google, DeepMind, and Facebook AI Research.\n- **Box upload:** 2026-05-06 14:56:56 PT — Day 8 mid-afternoon defense batch.\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~3.0 MB, 4 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `0724.pdf`.\n\n## Transcribed text\n\n> **From:** Shivon Zilis <shivon@tesla.com>\n> **To:** Sarah O'Brien <sobrien@tesla.com>\n> **Subject:** Re: Tesla AI Event (First Draft of FAQs)\n> **Date:** 11/27/2017 3:02:23 PM\n>\n> Happy to if you think helpful! Neuralink in morning then DC rest of day\n\n> **From:** Sarah O'Brien <sobrien@tesla.com>\n> **To:** Shivon Zilis <shivon@tesla.com>\n> **Subject:** Re: Tesla AI Event (First Draft of FAQs)\n> **Date:** Nov 27, 2017, at 2:45 PM\n>\n> Perfect. Keely and I are also meeting with Andrej tomorrow to talk about comms/tweets/blog posts etc. Wanna join? Or I'll fill you in afterwards.\n\n> **From:** Shivon Zilis <shivon@tesla.com>\n> **To:** Sarah O'Brien <sobrien@tesla.com>\n> **Subject:** Re: Tesla AI Event (First Draft of FAQs)\n> **Date:** Nov 27, 2017, at 2:39 PM\n>\n> Thanks for the feedback Sarah! I'm adding Andrej and Aimee here to keep them in the loop on the back and forth on content here!\n\n> **From:** Sarah O'Brien <sobrien@tesla.com>\n> **To:** Shivon Zilis <shivon@tesla.com>\n> **Cc:** Sam Teller <steller@tesla.com>, Keely Sulprizio <ksulprizio@tesla.com>, Jenna Lutwin <JLutwin@tesla.com>\n> **Subject:** Re: Tesla AI Event (First Draft of FAQs)\n> **Date:** Monday, November 27, 2017 at 2:29 PM\n>\n> This is really helpful Shivon. Some thoughts/suggestions in red below.\n>\n> Looks like we have a meeting with Elon tomorrow afternoon on this. Jenna - it would be great to have venue recommendations before then if possible.\n>\n> Thanks\n>\n> Sarah\n\n> **From:** Shivon Zilis <shivon@tesla.com>\n> **To:** Sarah O'Brien <sobrien@tesla.com>\n> **Cc:** Sam Teller <steller@tesla.com>, Keely Sulprizio <ksulprizio@tesla.com>, Jenna Lutwin <JLutwin@tesla.com>\n> **Subject:** Tesla AI Event (First Draft of FAQs)\n> **Date:** Nov 27, 2017, at 1:01 AM\n>\n> Per our call, here is a first stab at an FAQ for thinking through the Tesla AI event. Please let me know how I can refine it to make it more helpful to you. As you can see, there are still a handful of unanswered questions that we will need to get guidance on.\n>\n> **THE EVENT**\n>\n> **What is this event?**\n> The purpose of this event is to share that **Tesla is building a world leading artificial intelligence lab** which will rival the likes of Google / DeepMind and Facebook AI Research. The attendees of this event will be AI researchers and engineers who are attending the annual NIPS conference. Attendance is by invitation only though, in addition to those we invite, attendees of the conference will be able to apply to attend. It will feature a fireside chat with Elon, Jim Keller, and Andrej Karpathy, followed by casual drinks and mingling.\n>\n> **Why share this message?**\n> Tesla is currently known for work in AI, but its AI street cred is entirely within the realms of applied AI for autonomous vehicles. **Elon's name is also associated with artificial intelligence more broadly, but that is primarily through OpenAI (one major issue for Tesla is when people think of Elon and AI, they think of OpenAI).** Those who want to work on large scale AI research don't currently think of Tesla, and **Elon wants to change that by announcing his intention to create a world class AI lab**.\n>\n> **What would be a successful outcome of this event?**\n> Convincing the top AI researchers and engineers that generally apply to Google Brain, Facebook AI Research (FAIR), and DeepMind that they should apply to Tesla as well -- and hopefully ultimately choose it over the others! Also, for those that are already working at these places, convincing them that **Tesla is a better home for them**.\n>\n> **TESLA AI**\n>\n> **What is Tesla AI?**\n> Tesla AI's goal is to bring about the best possible AI future. <<Need to figure out how much to disclose here… some of the strategy had yet to be defined and some of it may be deeply proprietary. The challenge here will be figuring out a message that's appealing to the group that's sharable.>> [O'Brien edit, in red:] This is the question that we'll need a really solid answer to, otherwise this will feel like a hollow moment.\n>\n> **Why do this?**\n> Unless something changes significantly in the world, we will usher in an era of artificial superintelligence. Furthermore, it will likely happen during our lifetimes. While we cannot know with certainty what will contribute to the safest AI future, we believe the following are directionally correct:\n>\n> a. **Ensuring a healthy balance of powers** — Currently, power is consolidated fairly exclusively within Google / DeepMind. It seems unlikely that the best future will come about with one horse being in control of the entire race. If one entity has a massive lead on creating superintelligence, it will be difficult to have checks and balances on how it evolves and affects the world.\n>\n> b. **Linking humanity's will with artificial intelligence** — It's not just about creating artificial superintelligence before everyone else, it's also about ensuring the best possible type of superintelligence is created. The more we can build safeguards, instill our ethics, and keep the algorithms aligned with our goals, the more we maximize the probability of a good outcome for humanity.\n>\n> a. **Frontloading regulation** — Since this will likely be the most powerful technology humanity has ever created, it stands to reason it should be closely monitored and regulated. Tesla has a history of working closely with regulators on products and could potentially lead the way for showing governments how to think about regulating AI effectively. The biggest risk here is regulation will happen too late, so the more Tesla AI can accelerate this, the better.\n>\n> **Why Tesla?**\n> Tesla is a natural home for building a world leading AI lab because:\n>\n> a. **Sustainability** — Tesla's mission has always been focus around the idea of sustainability. … It will double down on those capabilities to work towards ensuring the most sustainable AI future as well.\n>\n> b. **Hardware + software combination** — <<How much to disclose here?>>\n>\n> **Who?** Jim Keller, Andrej Karpathy, Elon Musk, Sam Altman?? <<TBD on descriptions… need to figure this out>>\n>\n> ---\n>\n> **GRUMPY FAQS**\n>\n> **If Elon is so scared of AGI, why is he rushing to build it?**\n> Elon does speak a lot about the dangers of artificial intelligence, once even referring to it as \"summoning the demon\" so messaging here needs to be done carefully. The true answer is we cannot guarantee a good AI future, but that it seems correct to have many parties building it with as much transparency and regulation as possible. The most used analogy here is nuclear weapons… would you rather have one nation having nuclear capabilities, or many? Most people would answer many, so there is a balance of powers and they keep each other in check.\n>\n> **Why would someone want to work at Tesla AI over Google Brain, DeepMind, and Facebook?**\n> <<TBD depending on what we are willing to disclose>> Possible answers include HW <-> SW advantage, the fact that Tesla's goal has always been putting humanity's best interests first, that this is the best place for anyone wanting to work on a truly large scale AI problem (vs. individual research, which is a better fit at OpenAI) [O'Brien: this is a good point I think], and for those that want the flexibility to work on applied AI like autopilot and robotic automation in addition to large scale general AI research.\n>\n> **How many people? Who will lead the effort? Is this separate from or linked with autopilot?** <<All TBD>>\n>\n> **Will this deliver value to the rest of Tesla? How will it produce shareholder value?** <<Not sure how you'd want to address this?>>\n>\n> **LOGISTICAL CONSIDERATIONS**\n>\n> **Moderator Ideas** (there are no bad ideas in brainstorming…)\n> - No moderator? (Just the three of them talking about the future?)\n> - **Sam Altman?** (to show non-adversarial nature of the OpenAI / TeslaAI relationship? — could be a forcing function for Sam to commit to Tesla AI as well) [O'Brien: I think a moderator would be good. Part of what we/Tesla/Elon need to do is bring excitement to the conversation. I think it's easier to do that with the right moderator.]\n> - An author? (Kevin Kelly? John Markoff? Erik Brynjolfsson?)\n>\n> **Journalists** (if we choose to invite any…) — Cade Metz, Tom Simonite, <<Or others that are already trusted?>>\n>\n> **Invitations** — Starter list to be collected from Andrej Karpathy (invite list with ability to bring a researcher who is a friend). Some sort of application page? E mentioned he may tweet about it a few days in advance.\n\n## Commentary\n\nDX 724 is one of defense's strongest **\"parallel AGI play\"** exhibits — but unlike the [[DX-619|Karpathy poach texts]], the contents are an explicit, written *strategic-communications* document. In November 2017 — only weeks after the [[PX 157|\"final straw\" email]] and the [[PX 151|Brockman August 21 journal entry]] — Musk's office (Zilis) was drafting public-facing FAQs for a Tesla AI event whose stated purpose was to **rival** Google, DeepMind, and Facebook AI Research, to \"**convince [researchers] Tesla is a better home for them**\" than OpenAI, and to anticipate the \"**If Elon is so scared of AGI, why is he rushing to build it?**\" question. The \"**individual research, which is a better fit at OpenAI**\" line is a tell: in Tesla-internal framing, OpenAI was being positioned as the *small-research-group* alternative to Tesla AI's *large-scale* effort. The \"**Sam Altman?**\" moderator suggestion (\"could be a forcing function for Sam to commit to Tesla AI as well\") is also striking — a Tesla-side coordination memo discussing using a public event to recruit OpenAI's CEO to Tesla. Defense uses this to undermine the \"stole-a-charity\" narrative: Musk treated Tesla AI and OpenAI as parallel, somewhat-rivalrous AGI ventures from at least November 2017, and his \"fool-providing-free-funding\" framing dovetails with rather than contradicts that. Cross-reference [[Key Themes]] § \"Tesla / xAI as Musk's own AGI play\" and § \"Day 8 — what Zilis added.\"\n\n---\n*See also:* [[Day 8|Day 8 digest]] · [[Shivon Zilis]] · [[Andrej Karpathy]] · [[Tesla AI]] · [[DX-619]] · [[DX-757]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-728", "exhibit": "DX 728", "party": "Defense", "type": "Email thread", "admitted_trial_day": "Day 8 (May 6, 2026)", "uploaded_box_pt": "2026-05-06T14:56:55", "uploader": "Morrison Foerster", "pages": 6, "size_bytes": 698647, "source_pdf": "DX-728.pdf", "pdf_url": "https://media.mts-in.com/DX-728.pdf", "body_markdown": "# DX 728 — Dec 21–22, 2017 OpenAI biweekly update + Teller commentary on Musk's Tesla focus\n\n> Teller-to-Zilis commentary on Sutskever's Dec 21, 2017 biweekly board update — including Teller's contemporaneous statement that \"**E said again yesterday that his overall biggest concern remains AI stuff and that once we are stable with Model 3 he really plans to ratchet up time allocation. Reiterated that he thinks very low probability of OpenAI succeeding at AGI and we need to build up Tesla**.\"\n\n## Document type\n**Email thread, 6 pages,** consisting of (i) a top message from Sam Teller <steller@openai.com> to Shivon Zilis <shivon@openai.com>, sent Fri 22 Dec 2017 11:58:54 PM, subject \"Re: Update\"; (ii) Zilis's earlier reply; and (iii) Sutskever's biweekly OpenAI status email of Dec 21, 2017 10:43 AM PST forwarded into the chain (recipients: Musk, Zilis, Teller; cc Brockman, Altman). Production-stamped CONFIDENTIAL; Bates OPENAI_MUSK00024218–00024223. Document ID DX-0728. Pages 3–6 are example paragraphs from a then-novel OpenAI language model trained on books — the exhibit reproduces them in full as part of the Sutskever update.\n\n## Logistics\n- **Trial admission:** Day 8 (May 6, 2026) — Murati live; Zilis live; Toner video begins. Used during Zilis's live testimony as part of the defense's \"Musk was already pivoting to Tesla\" thread that anchors the [[Key Themes#Tesla / xAI as Musk's own AGI play (the parallel)|\"parallel AGI play\"]] theme — specifically the December 2017 / February 2018 progression that ends in DX 758, DX 761 (\"actively try to move three or four people from OpenAI to Tesla\"), and DX 853 (\"zero percent\").\n- **Box upload:** 2026-05-06 14:56:55 PT — Day 8 mid-afternoon batch (clustered with DX 643, DX 707, DX 758, DX 925).\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~682 KB, 6 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `0728.pdf`.\n\n## Transcribed text\n\n> **From:** Sam Teller <steller@openai.com>\n> **To:** Shivon Zilis <shivon@openai.com>\n> **Subject:** Re: Update\n> **Date:** 12/22/2017 11:58:54 PM\n>\n> E said again yesterday that his overall biggest concern remains AI stuff and that once we are stable with Model 3 he really plans to ratchet up time allocation. Reiterated that he thinks very low probability of OpenAI succeeding at AGI and we need to build up Tesla, but as you've said there's a lot of detail missing from the picture. Fine to forward this to him and maybe say Looks like Reid has finally started contributing. These guys have been slightly confused and unsure of your plans since the Tesla AI event and so a \"Good update\" reply or acknowledgment from you to close out the year on a positive note would be well received...\n>\n> Something like that :)\n\n> **From:** Shivon Zilis <shivon@openai.com>\n> **To:** Sam Teller <steller@openai.com>\n> **Subject:** Re: Update\n> **Date:** Dec 22, 2017, at 4:04 PM\n>\n> FYI had a brief convo earlier this week with Altman about there being zero communication recently and he said he'd try to get things back on track.\n>\n> I actually think this is a great and honest summary. I'm not sure if Elon wants in or out on OpenAI, but may be good for him to acknowledge if we are keeping things open?\n>\n> Sorry if that's a silly comment. Just know that it took Greg and Ilya a bit of nudging given they felt shitty about their candidate getting unsold by E.\n>\n> Also — FYI, they are trying super hard to source and close badass female engineers (easier than ML researchers) so they are moving on that front.\n>\n> Begin forwarded message:\n\n> **From:** Ilya Sutskever <ilyasu@openai.com>\n> **To:** Elon Musk <erm@spacex.com>, Shivon Zilis <shivon@openai.com>, Sam Teller <steller@spacex.com>\n> **Cc:** Greg Brockman <gdb@openai.com>, Sam Altman <shga@ycombinator.com>\n> **Subject:** Update\n> **Date:** December 21, 2017 at 10:43:22 AM PST\n>\n> Going forward, we will resume sending these emails every 2 weeks.\n>\n> **Dota:**\n> - *Finally* received more GPUs from Azure (1,000 Pascal GPUs). They are exposed by Azure in a hard-to-use way, but we'll have them up and running shortly.\n> - Fixed many 5v5 bugs, now showing strategic signs of life (split lanes, heroes roam, last hits/denies): [YouTube link]\n> - Goals:\n>   - [Jan 31] Reproduce superhuman 1v1 shadowfiend result. Valve released a major patch which broke some of our code. Needed the new compute to arrive to rerun the 1v1 training without disrupting rest of company, and also removed hacks so code usable for 5v5. Scoreboard: [Dropbox link]\n>   - [Feb 28] Superhuman 1v1 with any of 10 heroes.\n>   - [April] Beat pros in maximally restricted 5v5 (mirror match, turbo mode, fixed 5 heroes)\n>   - [June-August] Beat pros in unrestricted 5v5 across 64 heroes\n>\n> **Robotics:**\n> - Consistent rotation of the locked cube: [YouTube]\n>   - Policy is now robust, makes sensible use of fingers\n> - Much more accurate state estimation using phase space trackers. (Previous trackers were off by 1cm.)\n> - Early results on vision-based state estimation, will use for end-to-end training of vision-to-action model\n>   - Using machine learning to make simulator images look just like reality: [Dropbox]\n> - Goal: [Feb 15] Hand solving Rubik's cube\n>\n> **Data center preparation:**\n> - Machines from Shane: about 2x slower on our benchmark than our current build (partly explained by DMA, but the gap is surprising)\n> - RDMA over Ethernet working well: ~2us latency and 97Gbit/s throughput between two servers with full offload to the NIC — no need for InfiniBand\n> - Tested assembling servers (1-2 hour per, could probably get to 30 mins), believe we should outsource this — e.g. FB outsources to an ODM like Quanta\n> - Currently testing IPMI server management; Ceph for distributed storage\n> - Ready to start next buildout phase, exact plans depend on cloud vendor decisions + funding commitments\n>\n> **Other:**\n> - [Jan] Releasing public contest for designing algorithms that new learn games rapidly\n>   - Useful for recruiting and advancing our algorithms\n> - A neural network that can generate text with good coherence across multiple paragraphs [three examples at the end of the email]\n> - Met with Henry Kissinger to discuss AI strategy (he's made AI a focus area since hearing Demis speak at Bilderberg), he wants to keep meeting\n> - OpenAI now has person working on long-term strategy full-time, Michael Page\n> - **Fundraising:** people find our technical case convincing, we're still weak on why not just Google build it or why a non-profit is necessary (though some believe this on their own). Nevertheless, looks like we can raise plenty of short-term capital (especially after we complete the hand). With **$100M committed over the year, we can build our first data center.**\n>   - Reid has started donating, Nicolas Berggruen ready to\n>   - Active convos with Herb Allen, Jaan Tallin, Vinod Khosla\n>   - John Elkann wants to donate robots, but not funds (working with Kissinger to set up a policy initiative and doesn't want a conflict of building)\n> - Learned that China's government-linked AI results are not published publicly, but instead circulated in-country on private distribution lists. Hard to know how advanced China is (they seem to win the contests these days).\n>\n> **Challenges:**\n> - At NIPS, our people heard consistent negative feedback on: firings and Pieter/students leaving OpenAI, not publishing the inner workings of Dota, diversity.\n> - Have been interviewing many OpenAI folk to gauge morale, which has surfaced low-hanging organizational/management fruit:\n>   - Recruiting candidate experience, regular manager-IC meetings, regular feedback, better cross-team communications, etc\n>   - Addressing urgently, expect to have addressed outstanding issues by end of January\n>\n> *[Pages 3–6 reproduce three illustrative paragraphs of model-generated prose from a then-experimental OpenAI language model \"trained on books\" — sample dialogue passages used to demonstrate multi-paragraph coherence. Sutskever's framing: \"Once the generated samples make perfect sense 100% of the time, true language understanding will be around the corner.\" Sample text omitted as non-substantive.]*\n\n## Commentary\n\nDX 728 lands two of the defense's most useful December-1917-vintage facts. **First**, Teller — Musk's chief of staff — telling Zilis in writing that Musk *himself* was saying \"**very low probability of OpenAI succeeding at AGI and we need to build up Tesla**\" in December 2017 anticipates by exactly one year the Dec 31, 2018 [[DX-853]] \"**zero percent**\" email and is the best contemporaneous evidence the defense has for the [[Key Themes#Tesla / xAI as Musk's own AGI play (the parallel)|Tesla-as-AGI-play]] theme. **Second**, Sutskever's biweekly status report — the kind of routine, granular operational update Musk was receiving as a board member — undercuts any suggestion that Musk was being kept in the dark about OpenAI's plans, fundraising pipeline (Reid Hoffman, Berggruen, Khosla, \"$100M committed over the year\"), Azure compute dependence, and headcount/morale issues. The fundraising bullet — \"still weak on why not just Google build it or why a non-profit is necessary\" — is the kind of internal-doubt language plaintiffs will read as evidence the founders were already drifting toward the for-profit conversion plaintiffs say crystallized in 2018–2019 ([[DX-827]] purple box). The Kissinger meeting and the China-publication gap will be light color. The model-generated text on pages 3–6 is the public-facing version of what would, in 2018, become GPT-1 — included in the Sutskever update to demonstrate progress, not for any litigation purpose.\n\n---\n*See also:* [[Day 8|Day 8 digest]] · [[Sam Teller]] · [[Shivon Zilis]] · [[Ilya Sutskever]] · [[DX-643]] · [[DX-758]] · [[DX-853]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-734", "exhibit": "DX 734", "party": "Defense", "type": "Email", "admitted_trial_day": "Day 7 (May 5, 2026)", "uploaded_box_pt": "2026-05-05T14:27:00", "uploader": "Morrison Foerster", "pages": 2, "size_bytes": 445225, "source_pdf": "DX-734.pdf", "pdf_url": "https://media.mts-in.com/DX-734.pdf", "body_markdown": "# DX-734 — Jan 8, 2018 Brockman biweekly update: AGI weekend funder pipeline\n\n> Greg Brockman's biweekly update emailed to Musk, Sutskever, and Altman — naming the funders he's lining up for \"AGI weekend\" (Hoffman, Berggruen, Tallinn, McCaleb, Jensen, Newell) and laying out a $50M-by-end-of-January / $100M-by-end-of-Q1 fundraising goal.\n\n## Document type\n**Email, plain text bullet-list update.** From Greg Brockman <gdb@openai.com> to Elon Musk <erm@openai.com>; cc Ilya Sutskever, Sam Altman. Subject: \"biweekly update,\" Mon, 8 Jan 2018 23:15:15 -0800. Bates 2024MUSK-0005513–0005514.\n\n## Logistics\n- **Trial admission:** Day 7 (May 5, 2026) — Brockman defense direct + Mr. Molo recross + DX 1252; Wu deposition.\n- **Box upload:** 2026-05-05 14:27:00 PT — same Day-7 batch as DX 622.\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~445 KB, 2 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `0734.pdf`.\n\n## Transcribed text\n\n> **From:** Greg Brockman <gdb@openai.com>\n> **To:** Elon Musk <erm@openai.com>\n> **Cc:** Ilya Sutskever <ilyasu@openai.com>, Sam Altman <shga@ycombinator.com>\n> **Subject:** biweekly update\n> **Date:** Mon, 8 Jan 2018 23:15:15 -0800\n>\n> *Ask:* During AGI weekend, help close the funders listed below, and entice other attendees (e.g. Khosla, Milner — we've pitched Khosla and he's interested, haven't pitched Milner).\n>\n> ***Fundraising***\n>\n> - Most advanced funding conversations below (* = attending AGI weekend):\n>   - Reid Hoffman (*): *committed* for $20M/yr (has wired first quarterly installment). Discussing a larger amount.\n>   - Nicolas Berggruen (*): considering $20M donation, will decide after AGI conference. **Decision criteria:** whether he wants to be involved in building AGI directly, rather than just policy through the institute.\n>   - Jaan Tallin (*): considering a $20M donation. **Decision criteria:** whether we're thinking about safety + strategy properly\n>   - Jed McCaleb (*): has 2B Ripple (right now worth $2.36 each) in a donor advised fund, considering committing up to half.\n>   - Greg Jensen: considering $10M a year for 3 years. Would have closed in 2017 but he got caught up in this: https://www.wsj.com/articles/bridgewater-paid-over-1-million-to-employee-pushed-out-after-relationship-with-dalios-right-hand-man-1510050781. Will need to decide if we still want him.\n>   - Gabe Newell: considering a $20M donation. Met him today, he said he's interested and going to think about it for a day. Asked to join an advisory board.\n> - Funding situation:\n>   - Current 2018 budget: $5M from Sam Altman, $10M from Open Philanthropy; $20M from Reid, O($100M) donated compute from GCP\n>   - Lower bound goal: total $50M in 2018 cash by end of January, $100M in 2018 cash by end of Q1\n>\n> Organizational:\n> - Research manager, David Luan, started. Currently working on recruiting and internal processes.\n> - Greg and Ilya working on clearing up organizational debt\n>\n> ***Technical***\n>\n> Robotics:\n> - Progress on visual state estimation (links to YouTube and Drive)\n> - Train by automatically generating realistic input images from simulated images using CycleGAN\n> - Automatically adjusting the randomization parameters of the simulation to reduce the difficulty of the learning problem\n> - Will have video by end of weekend\n>\n> Dota:\n> - On track to reproducing our 1v1 Shadow Fiend results this week. (TrueSkill of the running experiment screenshot.)\n> - As a reminder, our goal is to reproduce Shadow Fiend 1v1 results by Jan 31, and superhuman with any of 10 heroes by end of Feb\n>\n> Games:\n> - Public contest for rapid learning of previously-unseen games to be released in early Feb. Goal is to recruit unknown talent, PR within community\n>\n> Other:\n> - Progress on Montezuma's Revenge via curiosity-based exploration (link). (expect to push this to beat DeepMind's best results soon, and has a 50% chance of completely solving)\n\n## Commentary\n\nDX 734 is the defense's chronological bridge between the Sept 2017 control fight and Musk's actual departure: by January 2018 OpenAI is running a real nonprofit funding pipeline — Reid Hoffman wired-in for $20M/yr, $5M from Altman, Open Philanthropy at $10M, GCP donating ~$100M of compute — and Musk himself remains the addressee of the biweekly update at his openai.com address. This contradicts the plaintiffs' \"they couldn't survive without me\" framing and supplies the [[Day 7|Day 7]] Brockman defense direct with concrete numbers. It is also where Gabe Newell first appears in the narrative — Musk's November 2018 follow-up email to Newell (\"**attempt that through Tesla instead**\") is in the [[Key Themes]] [[DX-844]] arc. The email is light on for-profit conversion talk because that fight had been \"tabled\" per [[DX-708]], but the Reid Hoffman / Berggruen / Tallinn / McCaleb / Newell pipeline framed in donation terms supports the defense narrative that Musk *did* know OpenAI was a charitable solicitation operation in 2018.\n\n---\n\n*See also:* [[Day 7|Day 7 digest]] · [[Greg Brockman]] · [[DX-708]] · [[DX-622]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-748", "exhibit": "DX 748", "party": "Defense", "type": "Email", "admitted_trial_day": "Day 7 (May 5, 2026)", "uploaded_box_pt": "2026-05-05T14:27:00", "uploader": "Morrison Foerster", "pages": 5, "size_bytes": 1100627, "source_pdf": "DX-748.pdf", "pdf_url": "https://media.mts-in.com/DX-748.pdf", "body_markdown": "# DX 748 — Feb 1, 2018 Brockman → Musk \"Top AI institutions today\" / \"Moral high ground\" reply\n\n> Brockman's long-form answer to Musk's January 31, 2018 ultimatum (\"**OpenAI is on a path of certain failure relative to Google**\") — laying out a fundraising plan, a four-headcount-year ramp (40 → 900), and the \"**moral high ground**\" doctrine: \"**Try our best to remain a non-profit**\" because \"our fiduciary duty should be to humanity.\"\n\n## Document type\n**Email thread, five pages, four messages.** Brockman's February 1, 2018 10:55 AM PST reply on top of Musk's January 31, 2018 4:04 AM \"two thumbs up\" reply, atop Brockman's January 31, 2018 10:57 PM \"Hi Elon — Thank you for the thoughtful note\" original, atop the originating January 31, 2018 1:20 PM Andrej Karpathy (akarpathy@tesla.com) → Musk note \"Top AI institutions today\" attaching the ICLR papers-by-institution chart. Recipients: Musk, Sutskever, Altman, Sam Teller, Shivon Zilis. Subject \"Re: Top AI institutions today.\" Bates 2024MUSK-0004547–4551. Document ID DX-0748.\n\n## Logistics\n- **Trial admission:** Day 7 (May 5, 2026) — used during the Brockman defense direct to anchor the \"moral high ground\" mission language at the moment Musk was simultaneously pulling Karpathy onto Tesla AI ([[DX 766]] companion).\n- **Box upload:** 2026-05-05 14:27:00 PT — Day 7 mid-afternoon batch (clustered with DX 626).\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~1.07 MB, 5 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `0748.pdf`.\n\n## Transcribed text\n\n> **From:** Greg Brockman <gdb@openai.com>\n> **To:** Elon Musk <erm@spacex.com>\n> **Cc:** \"ilyasu@openai.com\" <ilyasu@openai.com>, Sam Altman <shga@ycombinator.com>, Sam Teller <steller@spacex.com>, Shivon Zilis <Shivon@spacex.com>\n> **Subject:** Re: Top AI institutions today\n> **Date:** Thu, 1 Feb 2018 10:55:31 -0800\n> **Inline-Images:** pastedImage.png\n>\n> It seems there are several questions we must answer:\n>\n> 1. How much money do we think will be required to get to the final AGI? (And with what ramp?)\n> 2. How much certainty do we need today on our ability to raise each stage of that ramp? (Our ability to fundraise today is already much better than a year ago, and I expect e.g. US government funding to become viable in the future.)\n> 3. What ramp of people are needed, and doing what activities? (Would be curious to get Andrej's read, but my perspective is that working on a self-driving car is not vector aligned with working on AGI, and many of the best people don't want to work on self-driving cars.)\n> 4. How far do we want to go to pull/attract talent from Google/DM? (You'll need something very special and differentiated — high cash salary isn't enough — to pull people away.)\n> 5. How do we want to benchmark ourselves next to the competition? (Paper count? Number of smart people? Big demonstrations? \"Know it when I see it\"?)\n>\n> I feel we must resolve these questions before taking on new board members. (Of course, we can invite our candidate board members to partake in this conversation.)\n>\n> - gdb\n\n> **From:** Elon Musk <erm@spacex.com>\n> **To:** Greg Brockman <gdb@openai.com>\n> **Cc:** Ilya Sutskever <ilyasu@openai.com>, Sam Altman <shga@ycombinator.com>, Sam Teller <steller@spacex.com>, Shivon Zilis <shivon@spacex.com>\n> **Subject:** Re: Top AI institutions today\n> **Date:** Thu, Feb 1, 2018 at 4:04 AM\n>\n> Reid, Gabe and Adam are two thumbs up from me. Don't know the others well enough yet.\n>\n> I'm still really concerned that we won't be able to grow fast enough. Frankly, unless we are able to pull significant numbers of key people from Google/Deepmind, we should assume failure.\n>\n> It feels like we are similar to Blue Origin. Bezos is still clueless as to how hopelessly far behind he is and constantly rationalizes his position, overestimating his ability and dramatically underestimating SpaceX year after year.\n>\n> We lost an excellent engineer to BO last year (first one in several years) and were very concerned. That guy recently returned to SpaceX and told everyone how hopeless they were. Now zero people are even taking BO recruiting calls. Doesn't matter what they offer.\n\n> **From:** Greg Brockman <gdb@openai.com>\n> **To:** Elon Musk <erm@spacex.com>\n> **Cc:** Ilya Sutskever <ilyasu@openai.com>, Sam Altman <shga@ycombinator.com>, Sam Teller <steller@spacex.com>, Shivon Zilis <shivon@spacex.com>\n> **Subject:** Re: Top AI institutions today\n> **Date:** Jan 31, 2018, at 10:57 PM\n>\n> Hi Elon,\n>\n> Thank you for the thoughtful note. I have always been impressed by your focus on the big picture, and agree completely we must change trajectory to achieve our goals. Let's speak tomorrow, any time 4p or later will work.\n>\n> My view is that the best future will come from a major expansion of OpenAI. Our goal and mission are fundamentally correct, and that will increasingly be a superpower as AGI grows near.\n>\n> **# Fundraising**\n>\n> Our fundraising conversations show that:\n>\n> - Ilya and I are able to convince reputable people that AGI can really happen in the next <=10 years\n> - There's appetite for donations from those people\n> - There's *very* large appetite for investments from those people\n>\n> I respect your decision on the ICO idea, which matches the evolution of our own thinking. Sam Altman has been working on a fundraising structure that does not rely on a public offering, and we will be curious to hear your feedback.\n>\n> Of the people we've been talking to, the following people are currently my top suggestions for board members. Would also love suggestions for your top picks not on this list, and we can figure out how to approach them.\n>\n> - Reid Hoffman\n> - Gabe Newell\n> - Adam d'Angelo\n> - Jed McCaleb\n> - Herb Allen\n> - Terah Lyons (she heads Partnership on AI, originally created by Demis to steal OpenAI's thunder — would bring a lot of outside credibility)\n>\n> **# The next 3 years**\n>\n> Over the next 3 years, we must build 3 things:\n>\n> - Custom AI hardware (such as Jim's computer)\n> - Massive AI data center (likely multiple revs thereof)\n> - Best software team, mixing between algorithm development, public demonstrations, and safety\n>\n> We've talked the most about the custom AI hardware and AI data center. On the software front, we have a credible path (self-play in a competitive multiagent environment) which has been validated by Dota and AlphaGo. We also have identified a small but finite number of limitations in today's deep learning which are barriers to learning from human levels of experience. And we believe we uniquely are on trajectory to solving safety (at least in broad strokes) in the next three years.\n>\n> We would like to scale headcount in this way:\n>\n> - Beginning of 2017: ~40\n> - End of 2018: 100\n> - End of 2019: 300\n> - End of 2020: 900\n>\n> Note that many of DeepMind's 600+ people are not working on AGI, but instead DeepMind Health or DeepMind for Google. Every single one of our people will be working on AGI.\n>\n> **## Moral high ground**\n>\n> Our biggest tool is the moral high ground. To retain this, we must:\n>\n> - **Try our best to remain a non-profit. AI is going to shake up the fabric of society, and our fiduciary duty should be to humanity.**\n> - Put increasing effort into the safety/control problem, rather than the fig leaf you've noted in other institutions. It doesn't matter who wins if everyone dies. Related to this, we need to communicate a \"better red than dead\" outlook — we're trying to build safe AGI, and we're not willing to destroy the world in a down-to-the-wire race to do so.\n> - Engage with government to provide trusted, unbiased policy advice — we often hear that they mistrust recommendations from companies such as Intel, Google, etc..\n> - Be perceived as a place that provides public good to the research community, and keeps the other actors honest and open via leading by example.\n>\n> **# The past 2 years**\n>\n> I would be curious to hear how you rate our execution over the past two years, relative to resources. In my view:\n>\n> - Over the past five years, there have two major demonstrations of working systems: AlphaZero [DeepMind] and Dota 1v1 [OpenAI]. (There are a larger number of breakthroughs of \"capabilities\" popular among practitioners, the top of which I'd say are: ProgressiveGAN [NVIDIA], unsupervised translation [Facebook], WaveNet [DeepMind], Atari/DQN [DeepMind], machine translation [Ilya at Google — now at OpenAI], generative adversarial network [Ian Goodfellow at grad school — now at Google], variational autoencoder [Durk at grad school — now at OpenAI], AlexNet [Ilya at grad school — now at OpenAI].) We benchmark well on this axis.\n> - We grew very rapidly in 2016, and in 2017 iterated to a working management structure. We're now ready to scale massively, given the resources. We lose people on comp currently, but pretty much *only* on comp. I've been resuming the style of recruiting I did in the early days, and believe I can exceed those results.\n> - We have the most talent dense team in the field, and we have the reputation for it as well.\n> - We don't encourage paper writing, and so paper acceptance isn't a measure we optimize. For the ICLR chart Andrej sent, I'd expect our (accepted papers)/(people submitting papers) to be the highest in the field.\n>\n> - gdb\n\n> **From:** Elon Musk <erm@spacex.com>\n> **To:** Greg Brockman <gdb@openai.com>\n> **Cc:** Ilya Sutskever <ilyasu@openai.com>, Sam Altman <shga@ycombinator.com>, Sam Teller <steller@spacex.com>, Shivon Zilis <shivon@spacex.com>\n> **Subject:** Re: Top AI institutions today\n> **Date:** Wed, Jan 31, 2018 at 2:02 PM\n>\n> OpenAI is on a path of certain failure relative to Google. There obviously needs to be immediate and dramatic action or everyone except for Google will be consigned to irrelevance.\n>\n> I have considered the ICO approach and will not support it. In my opinion, that would simply result in a massive loss of credibility for OpenAI and everyone associated with the ICO. If something seems too good to be true, it is. This was, in my opinion, an unwise diversion.\n>\n> The only paths I can think of are a major expansion of OpenAI and a major expansion of Tesla AI. Perhaps both simultaneously. **The current board situation is very weak.**\n>\n> I will set up a time for us to talk tomorrow. To be clear, I have a lot of respect for your abilities and accomplishments, but I am not happy with how things have been managed. That is why I have had trouble engaging with OpenAI in recent months. **Either we fix things and my engagement increases a lot or we don't and I will drop to near zero and publicly reduce my association.** I will not be in a situation where the perception of my influence and time doesn't match the reality.\n>\n> Begin forwarded message:\n\n> **From:** Andrej Karpathy <akarpathy@tesla.com>\n> **To:** Elon Musk <erm@tesla.com>\n> **Cc:** Shivon Zilis <shivon@tesla.com>\n> **Subject:** Top AI institutions today\n> **Date:** January 31, 2018 at 1:20:42 PM PST\n>\n> The ICLR conference (which is the top deep learning - specific conference (NIPS is larger, but more diffuse)) released their decisions for accepted/rejected papers, and someone made some nice plots that show where the current deep learning / AI research happens at. It's an imperfect measure because not every company might prioritize paper publications, but it's indicative.\n>\n> Here's a plot that shows the total number of papers (broken down by oral/poster/workshop/rejected) from any institution:\n>\n> [chart: ICLR papers by institution — google.com dominates with 83 submissions; academic institutions Berkeley/Stanford/CMU/MIT in 20-30 range; OpenAI also visible]\n>\n> Long story short, Google is dominating with 83 paper submissions. The academic institutions (Berkeley / Stanford / CMU / MIT) are next, in 20-30 ranges each.\n>\n> Just thought it was an interesting snapshot of where all the action is today. The full data is here: http://webia.lip6.fr/~pajot/dataviz.html\n>\n> -Andrej\n\n## Commentary\n\nDX 748 is the defense's headline mission-language exhibit. The \"**Moral high ground**\" section — \"**Try our best to remain a non-profit. AI is going to shake up the fabric of society, and our fiduciary duty should be to humanity**\" — is exactly the language that becomes the [[PX 24]] OpenAI Charter ten weeks later (April 9, 2018). Plaintiffs cite this passage to argue that the founders themselves committed to a non-profit fiduciary duty to humanity that the later for-profit conversion betrayed. Defense, however, frames the same email as Brockman *answering* a Musk ultimatum — \"**OpenAI is on a path of certain failure relative to Google**,\" \"**The current board situation is very weak**,\" and \"**I will drop to near zero and publicly reduce my association**\" — which the defense reads as Musk forcing the founders to choose a path forward without him. The originating Karpathy → Musk note is sent from `akarpathy@tesla.com` to `erm@tesla.com` — i.e., Karpathy was already a Tesla employee on Musk's Tesla email by January 31, 2018, two weeks before Musk's [[PX 157]]-aftermath board exit, supporting the defense's \"Tesla AI as Musk's parallel AGI play\" theme (see [[DX 766]]). Brockman's \"**working on a self-driving car is not vector aligned with working on AGI**\" line is the polite version of the in-house worry. Cross-reference: the headcount ramp (40 → 100 → 300 → 900) is the planning predicate for the [[DX 827]] LP raise that launched in August 2018.\n\n---\n*See also:* [[Day 7|Day 7 digest]] · [[Greg Brockman]] · [[PX 24]] · [[DX 766]] · [[DX 827]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-749", "exhibit": "DX 749", "party": "Defense", "type": "Email thread", "admitted_trial_day": "Day 3 (April 29, 2026)", "uploaded_box_pt": "2026-04-29T15:36:42", "uploader": "Morrison Foerster", "pages": 4, "size_bytes": 996126, "source_pdf": "DX-749.pdf", "pdf_url": "https://media.mts-in.com/DX-749.pdf", "body_markdown": "# DX 749 — Feb 1, 2018 Musk endorses Karpathy's \"OpenAI to attach to Tesla as its cash cow\" reply\n\n> The thread in which Andrej Karpathy (writing from `akarpathy@tesla.com`) tells Musk that OpenAI should \"**attach to Tesla as its cash cow**,\" and Musk forwards it on to Sutskever/Brockman/Altman with the all-caps endorsement: \"**Andrej is exactly right. We may wish it otherwise, but, in my and Andrej's opinion, Tesla is the only path that could even hope to hold a candle to Google.**\"\n\n## Document type\n**Email thread, plain text, four messages.** Page 1: Musk's Feb 1, 2018 11:52:14 AM (UTC) message to Sutskever/Brockman (\"Andrej is exactly right…\"); page 2: continuation of Karpathy's Jan 31, 2018 11:54:30 PM PST reply to Musk; page 2 also: Musk's Jan 31, 2018 2:07:15 PM forward to Karpathy (\"fyi What do you think makes sense? Happy to talk by phone if that's better.\") atop Musk's Jan 31, 2018 2:02:37 PM original to Sutskever, Brockman, Altman (\"OpenAI is on a path of certain failure relative to Google…\"); page 3: Karpathy's originating Jan 31, 2018 1:20:42 PM PST email to Musk attaching the ICLR institution chart; page 4: chart caption + sign-off. Subject \"Fwd: Top AI institutions today.\" Bates 2024MUSK-0005444–5447. Document ID DX-0749.\n\n## Logistics\n- **Trial admission:** Day 3 (April 29, 2026). Pre-trial wiki reference: **\"DX 749 — Karpathy attach-to-Tesla reply.\"** Per [[Key Themes#Tesla / xAI as Musk's own AGI play]]: \"Musk's '**OpenAI to attach to Tesla as its cash cow**' (DX 749, Karpathy reply Musk endorsed). `(042926TT.txt:4762)`.\"\n- **Box upload:** 2026-04-29 15:36:42 PT — Day 3 late-afternoon defense batch (clustered with [[DX 686]], [[DX 773]]).\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~973 KB, 4 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `749 .pdf` (note trailing space).\n\n## Transcribed text\n\n> **From:** \"Elon Musk\" <erm@spacex.com>\n> **To:** \"ilyasu@openai.com\", \"gdb@openai.com\"\n> **Subject:** Fwd: Top AI institutions today\n> **Date:** Thu, 01 Feb 2018 11:52:14 -0000\n> **Inline-Images:** pastedImage.png\n>\n> Andrej is exactly right. We may wish it otherwise, but, in my and Andrej's opinion, Tesla is the only path that could even hope to hold a candle to Google. Even then, the probability of being a counterweight to Google is small. It just isn't zero.\n>\n> Begin forwarded message:\n>\n> ---\n>\n> **From:** Andrej Karpathy <akarpathy@tesla.com>\n> **Date:** January 31, 2018 at 11:54:30 PM PST\n> **To:** Elon Musk <erm@spacex.com>\n> **Subject:** Re: Top AI institutions today\n>\n> > Working at the cutting edge of AI is unfortunately expensive. For example, DeepMind's operating expenses in 2016 were at around $250M USD (does not include compute). With their growing team today it might be ~0.5B/yr. But then Alphabet in 2016 reported ~20B net income so it's still fairly cheap even if DeepMind had no revenue of its own. In addition to DeepMind, Google also has Google Brain, Research, and Cloud. And TensorFlow, TPUs, and they own about a third of all research (in fact, they hold their own AI conferences).\n> >\n> > I also strongly suspect that compute horsepower will be necessary (and possibly even sufficient) to reach AGI. If historical trends are any indication, progress in AI is primarily driven by systems - compute, data, infrastructure. The core algorithms we use today have remained largely unchanged from the ~90s. Not only that, but any algorithmic advances published in a paper somewhere can be almost immediately re-implemented and incorporated. Conversely, algorithmic advances alone are inert without the scale to also make them scary.\n> >\n> > It seems to me that OpenAI today is burning cash and that the funding model cannot reach the scale to seriously compete with Google (an 800B company). If you can't seriously compete but continue to do research in open, you might in fact be making things worse and helping them out \"for free\", because any advances are fairly easy for them to copy and immediately incorporate, at scale.\n> >\n> > A for-profit pivot might create a more sustainable revenue stream over time and would, with the current team, likely bring in a lot of investment. However, building out a product from scratch would steal focus from AI research, it would take a long time and it's unclear if a company could \"catch up\" to Google scale, and the investors might exert too much pressure in the wrong directions.\n> >\n> > **The most promising option I can think of, as I mentioned earlier, would be for OpenAI to attach to Tesla as its cash cow.** I believe attachments to other large suspects (e.g. Apple? Amazon?) would fail due to an incompatible company DNA. Using a rocket analogy, Tesla already built the \"first stage\" of the rocket with the whole supply chain of Model 3 and its onboard computer and a persistent internet connection. The \"second stage\" would be a full self driving solution based on large-scale neural network training, which OpenAI expertise could significantly help accelerate. With a functioning full self-driving solution in ~2-3 years we could sell a lot of cars/trucks. If we do this really well, the transportation industry is large enough that we could increase Tesla's market cap to high O(~100K), and use that revenue to fund the AI work at the appropriate scale.\n> >\n> > I cannot see anything else that has the potential to reach sustainable Google-scale capital within a decade.\n> >\n> > -Andrej\n>\n> ---\n>\n> **From:** Elon Musk <erm@spacex.com>\n> **Sent:** Wednesday, January 31, 2018 2:07:15 PM\n> **To:** Andrej Karpathy\n> **Subject:** Fwd: Top AI institutions today\n>\n> > fyi\n> >\n> > What do you think makes sense? Happy to talk by phone if that's better.\n> >\n> > Begin forwarded message:\n> >\n> > ---\n> >\n> > **From:** Elon Musk <erm@spacex.com>\n> > **To:** Greg Brockman <gdb@openai.com>, Ilya Sutskever <ilyasu@openai.com>, Sam Altman <shga@ycombinator.com>\n> > **Cc:** Sam Teller <steller@spacex.com>, Shivon Zilis <Shivon@spacex.com>\n> > **Subject:** Fwd: Top AI institutions today\n> > **Date:** January 31, 2018 at 2:02:37 PM PST\n> >\n> > > OpenAI is on a path of certain failure relative to Google. There obviously needs to be immediate and dramatic action or everyone except for Google will be consigned to irrelevance.\n> > >\n> > > I have considered the ICO approach and will not support it. In my opinion, that would simply result in a massive loss of credibility for OpenAI and everyone associated with the ICO. If something seems too good to be true, it is. This was, in my opinion, an unwise diversion.\n> > >\n> > > The only paths I can think of are a major expansion of OpenAI and a major expansion of Tesla AI. Perhaps both simultaneously. The current board situation is very weak.\n> > >\n> > > I will set up a time for us to talk tomorrow. To be clear, I have a lot of respect for your abilities and accomplishments, but I am not happy with how things have been managed. That is why I have had trouble engaging with OpenAI in recent months. **Either we fix things and my engagement increases a lot or we don't and I will drop to near zero and publicly reduce my association.** I will not be in a situation where the perception of my influence and time doesn't match the reality.\n>\n> ---\n>\n> **From:** Andrej Karpathy <akarpathy@tesla.com>\n> **Date:** January 31, 2018 at 1:20:42 PM PST\n> **To:** Elon Musk <erm@tesla.com>\n> **Cc:** Shivon Zilis <shivon@tesla.com>\n> **Subject:** Top AI institutions today\n>\n> > The ICLR conference (which is the top deep learning - specific conference (NIPS is larger, but more diffuse)) released their decisions for accepted/rejected papers, and someone made some nice plots that show where the current deep learning / AI research happens at. It's an imperfect measure because not every company might prioritize paper publications, but it's indicative.\n> >\n> > Here's a plot that shows the total number of papers (broken down by oral/poster/workshop/rejected) from any institution:\n> >\n> > [chart: ICLR papers by institution — google.com dominates with 83 submissions; Berkeley/Stanford/CMU/MIT in the 20–30 range; Microsoft, Facebook, OpenAI also visible]\n> >\n> > Long story short, Google is dominating with 83 paper submissions. The academic institutions (Berkeley / Stanford / CMU / MIT) are next, in 20-30 ranges each.\n> >\n> > Just thought it was an interesting snapshot of where all the action is today. The full data is here: http://webia.lip6.fr/~pajot/dataviz.html\n> >\n> > -Andrej\n\n## Commentary\n\nDX 749 is the linchpin of the defense's [[Key Themes#Tesla / xAI as Musk's own AGI play (the parallel)|Tesla-as-Musk's-parallel-AGI-play]] theme. The Karpathy line — \"**OpenAI to attach to Tesla as its cash cow**\" — would be unremarkable if Karpathy were freelancing, but Musk **explicitly endorsed it** with \"**Andrej is exactly right**\" *and* forwarded the full thread to Sutskever/Brockman/Altman the same day with the parallel \"**major expansion of OpenAI and a major expansion of Tesla AI**\" pitch. Defense uses this for three independent points: (i) it disproves Musk's \"I had nothing to do with Tesla AI / OpenAI was my non-profit play\" framing — Karpathy was already on `akarpathy@tesla.com` while still doing OpenAI strategy work for Musk; (ii) the \"**increase Tesla's market cap to high O(~100K)** [billions]\" line is contemporaneous evidence that Musk was contemplating using OpenAI talent to *raise Tesla's valuation* — which is the personal financial conflict plaintiffs cannot easily explain away; and (iii) the embedded parallel email — \"**I will drop to near zero and publicly reduce my association**\" — is the conditional precursor to Musk's formal Feb. 2018 OpenAI board exit. Cross-reference: [[DX 748]] (Brockman's Feb. 1, 2018 \"moral high ground\" reply that *answers* the same Musk ultimatum) and [[DX 773]] (the June 2017 Keller-Karpathy \"OpenAI guys are gonna want to kill me\" thread).\n\n---\n*See also:* [[Day 3|Day 3 digest]] · [[Andrej Karpathy]] · [[DX 748]] · [[DX 773]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-753", "exhibit": "DX 753", "party": "Defendants", "type": "Text messages", "admitted_trial_day": "Day 8 (May 6, 2026)", "uploaded_box_pt": "2026-05-06T14:56:55", "uploader": "Morrison Foerster", "pages": 2, "size_bytes": 129898, "source_pdf": "DX-753.pdf", "pdf_url": "https://media.mts-in.com/DX-753.pdf", "body_markdown": "# DX 753 — Feb 3, 2018 Shivon Zilis-Sam Altman text exchange: \"did you think through B Corp subsidiary of Tesla?\"\n\n> A two-page extract from a 33-message text conversation between Shivon Zilis and Sam Altman dated February 3, 2018 — the day after Zilis sent Musk her \"bury this in Tesla for stealth advantage\" message (see [[Key Themes]] § \"'Bury this in Tesla for stealth advantage'\"). Zilis floats a \"B Corp subsidiary of Tesla\" structure to Altman; Altman responds: \"interesting idea — had not considered — will think.\" The exchange then turns to Falcon Heavy small talk.\n\n## Document type\n**Text messages.** A \"Short Message Report\" extract showing 33 messages on 2/3/2018 between two participants (Altman and Shivon Zilis); the exhibit reproduces only the first ~16 messages (covering 11:28 PM through 11:44 PM GMT). Bates ZILIS-0002089–2090.\n\n## Logistics\n- **Trial admission:** Day 8 (May 6, 2026) — Murati live; Zilis live; Toner video begins. Used by defense in connection with Zilis's live testimony — [[Key Themes]] § \"Day 8 — what Zilis added\" notes Schubert's effort to walk Zilis through the \"OpenAI possible scenarios\" planning (DX 758) and the \"burying\" theme; DX 753 is the contemporaneous companion text.\n- **Box upload:** 2026-05-06 14:56:55 PT — Day 8 mid-afternoon batch (uploaded 2 seconds after [[DX 715]]).\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~127 KB, 2 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `0753.pdf`.\n\n## Transcribed text\n\n> **Short Message Report**\n> Conversations: 1 · Total Messages: **33** · Participants: 2 · Date Range: 2/3/2018\n>\n> Conversation 97cf579ac2f61c9488dc8cce96ab8721 — 33 messages on 2/3/2018\n> Participants: Altman, Shivon Zilis\n>\n> ---\n>\n> **Messages in chronological order (times shown in GMT +00:00)**\n>\n> **Shivon Zilis** — 2/3/2018, 11:28 PM\n> Sorry if we talked about this — but did you think through **B Corp subsidiary of Tesla**?\n>\n> **Shivon Zilis** — 2/3/2018, 11:29 PM\n> Probably not the right answer, but it's the time to think through all possible ideas\n>\n> **Altman** — 2/3/2018, 11:30 PM\n> interesting idea\n>\n> **Altman** — 2/3/2018, 11:30 PM\n> had not considered\n>\n> **Altman** — 2/3/2018, 11:30 PM\n> will think\n>\n> **Altman** — 2/3/2018, 11:30 PM\n> you going to falcon heavy??\n>\n> **Shivon Zilis** — 2/3/2018, 11:31 PM\n> I so badly want to but feels too entitled 😞\n>\n> **Altman** — 2/3/2018, 11:31 PM\n> not at all\n>\n> **Altman** — 2/3/2018, 11:31 PM\n> it's your bday\n>\n> **Altman** — 2/3/2018, 11:31 PM\n> and historic\n>\n> **Shivon Zilis** — 2/3/2018, 11:32 PM\n> You def doing it?\n>\n> **Altman** — 2/3/2018, 11:35 PM\n> waiting for elissa to confirm i can\n>\n> **Altman** — 2/3/2018, 11:35 PM\n> if so yes\n>\n> **Altman** — 2/3/2018, 11:35 PM\n> big milestone in leaving the planet…\n>\n> **Shivon Zilis** — 2/3/2018, 11:43 PM\n> Oh it's not publicly viewable? Haha then no way I'm going. Would never in a million years ask!!!\n>\n> **Shivon Zilis** — 2/3/2018, 11:43 PM\n> I hope you have so much fun!!!\n>\n> **Shivon Zilis** — 2/3/2018, 11:44 PM\n> The simulation video alone gave me chills tens times over. How incredible\n\n[The exhibit indicates 33 total messages on 2/3/2018; the produced extract terminates at 11:44 PM. The remaining 17 messages are not included in the produced PDF.]\n\n## Commentary\n\nDX 753 is a small, contemporaneous document positioning the **B Corp subsidiary of Tesla** structure as a live February 2018 idea floated by Zilis to Altman directly — three days after Karpathy's \"OpenAI to attach to Tesla as its cash cow\" reply (DX 749, see [[Key Themes]] § \"Tesla / xAI as Musk's own AGI play\") and the day after Zilis's own \"burying\" email to Musk (DX 715 / DX 758 thread). **Defense's read:** the text shows that the \"Tesla absorbs OpenAI\" idea was being raised by *Zilis* (not Altman), and that Altman's response was a polite, non-committal \"interesting idea — had not considered — will think.\" Zilis is the live witness on Day 8 and a former plaintiff in this very case; her contemporaneous initiative on this topic undercuts plaintiffs' framing of the for-profit conversion as having been engineered by the founders to push Musk out. **Plaintiffs' read:** Zilis was Musk's intermediary and acting on his information needs — see her February 2018 \"we are going to actively try to move three or four people from OpenAI to Tesla\" exchange with Musk in DX 761 (cited at [[Key Themes]] § \"Tesla / xAI as Musk's own AGI play\"). The casual register of the Falcon Heavy small talk (\"**big milestone in leaving the planet**\" — Falcon Heavy's first launch was Feb 6, 2018) places these \"all possible ideas\" in a chatty insider context, not in a careful strategic-planning forum. The redacted second participant identifier (after \"Altman (-…)\") would correspond to Altman's personal phone number.\n\n---\n*See also:* [[Day 8|Day 8 digest]] · [[DX 715]] · [[DX 758]] · [[Shivon Zilis]] · [[Sam Altman]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-754", "exhibit": "DX 754", "party": "Defense (OpenAI/MS)", "type": "Email thread", "admitted_trial_day": "Day 8 (May 6, 2026)", "uploaded_box_pt": "2026-05-06T14:56:55", "uploader": "Morrison Foerster", "pages": 2, "size_bytes": 432723, "source_pdf": "DX-754.pdf", "pdf_url": "https://media.mts-in.com/DX-754.pdf", "body_markdown": "# DX 754 — Feb 3, 2018 Zilis / Teller \"So far\" debrief on Greg, Ilya, Altman, Karpathy\n\n> Shivon Zilis's February 3, 2018 readout to Sam Teller of separate conversations with Brockman, Sutskever, Altman, and Karpathy about whether the OpenAI principals would defect to Tesla — admitted on Day 8 with a limiting instruction.\n\n## Document type\n**Email thread, plain text, three messages on Feb 3, 2018** between Shivon Zilis and Sam Teller, subject \"Re: So far.\" Bates SPX-006856–006857. Production-stamped CONFIDENTIAL.\n\n## Logistics\n- **Trial admission:** Day 8 (May 6, 2026) — Murati live; Zilis live; Toner video begins. Per the wiki TSV: \"**DX 754 — limiting-instruction exhibit.**\" The thread was admitted with a limiting instruction (the out-of-court statements attributed to Brockman, Sutskever, Altman, and Karpathy come in for state-of-mind / effect on the listener, not for the truth of those non-declarants' assertions).\n- **Box upload:** 2026-05-06 14:56:55 PT — Day 8 mid-afternoon batch (clustered with [[DX-716]], [[DX-758]], DX 835, DX 877).\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~423 KB, 2 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `0754.pdf`.\n\n## Transcribed text\n\n> **From:** Shivon Zilis <shivon@spacex.com>\n> **To:** Sam Teller <steller@spacex.com>\n> **Subject:** Re: So far\n> **Date:** Sat, 3 Feb 2018 11:07:38 -0800\n>\n> Copy.\n>\n> I don't think it's just their egos, though that does play a part.\n>\n> I think they likely would come if E *really* did something. Like a real plan for how to build the counterbalance with some unique ideas. All they know right now is a vague promise of resources for this and they don't believe he has a real plan for how to build a counterbalance, which is true because he doesn't yet. I know if he thought through a plan it'd be better plan than anyone else's, but just hasn't happened yet. By plan I mean more vision / technical / ops than financial.\n>\n> Feels like he should come to that himself though, first — meaning of course with help if he wants — just not in collaboration with the two of them.\n>\n> Also, it's ok if they don't come and really fight to make OpenAI a real thing??\n\n> **From:** Sam Teller <steller@spacex.com>\n> **To:** Shivon Zilis <shivon@spacex.com>\n> **Subject:** Re: So far\n> **Date:** Feb 3, 2018, at 10:43 AM\n>\n> Copy that, thanks so much\n>\n> Hard to see Greg and Ilya's egos allowing them to come to Tesla\n>\n> E just suggested we use SpaceX email for AI stuff so switching over to that\n\n> **From:** Shivon Zilis <shivon@openai.com>\n> **To:** Sam Teller <steller@spacex.com>\n> **Subject:** So far\n> **Date:** Feb 3, 2018, at 10:18 AM\n>\n> All signs point to not-Tesla for Greg and Ilya. Altman seems most open and E could probably win him over with a real plan if he wanted to.\n>\n> Andrej's take was that Greg and Ilya were trying to be open minded enough to talk about what Tesla would be like but weren't really exploring it in earnest.\n>\n> Andrej cares a lot more about AGI than he does AI / computer vision applications. I didn't realize that until now. He is a little scared that two experiments may not be better than one (division of talent). Didn't fully get why E would want to separate from OpenAI but now seems to mostly understand.\n>\n> Other notes:\n> -Greg/Ilya/Altman said that they are not sharing this info with the team for now\n> -Altman still eager to collaborate closely with E even if separate entities (mentioned he wanted to ask him about some secdef stuff)\n> -I think a lot of how separation would go is predicated on whether they do ICO or not… if they don't, doesn't seem like a need to anyone to really know until E has a plan for Tesla AI and needs to be public about it\n> -Andrej thinks robotics a complete waste of time for AGI\n> -Greg had asked about working with E at Tesla. I'd shared about how I think E best in the world at accelerating engineering systems, so if path to AGI looks like DOTA / other environments in large team settings no better person to work with than E if you wanna get there quickly. That seemed to excite him but I still don't think Tesla truly an option for him yet.\n>\n> That's what I got!\n>\n> Shivon\n\n## Commentary\n\nDX 754 is the live operational counterpart to [[DX-716]] (Oct 2017) and the run-up to [[DX-758]] (Feb 13, 2018, ten days later): by early February 2018 Musk's emissaries were systematically canvassing the OpenAI principals about whether they would follow Musk to Tesla. Zilis's frank \"**E… doesn't [have a real plan] yet**\" and her note that Brockman had \"asked about working with E at Tesla\" support the defense [[Key Themes#Tesla / xAI as Musk's own AGI play (the parallel)|parallel-AGI-play]] frame and complement Karpathy's same-month \"**actively try to move three or four people from OpenAI to Tesla**\" exchange ([[DX 761]]). The limiting instruction matters: the jury was told it could not credit Brockman/Sutskever/Altman/Karpathy's reported views *as true* — only that Zilis (and through her, Musk) was operating on those impressions in February 2018. Plaintiffs will argue the thread shows the founders were *not* defecting and were \"really fight[ing] to make OpenAI a real thing\"; defense will argue it shows Musk was already preparing the Tesla alternative weeks before the [[DX-758|\"Possible Scenarios\"]] memo. See [[Day 8|Day 8 digest]] and [[Key Themes#\"Bury this in Tesla for stealth advantage\"]].\n\n---\n*See also:* [[Day 8|Day 8 digest]] · [[Shivon Zilis]] · [[DX-716]] · [[DX-758]] · [[DX 761]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-756", "exhibit": "DX 756", "party": "Defense", "type": "Email thread", "admitted_trial_day": "Day 8 (May 6, 2026)", "uploaded_box_pt": "2026-05-06T14:56:55", "uploader": "Morrison Foerster", "pages": 2, "size_bytes": 230114, "source_pdf": "DX-756.pdf", "pdf_url": "https://media.mts-in.com/DX-756.pdf", "body_markdown": "# DX 756 — Feb 8–9, 2018 Sam Teller ↔ Shivon Zilis thread, \"A few things\" (limiting-instruction exhibit)\n\n> A four-day Sam Teller ↔ Shivon Zilis email exchange (Feb 8–9, 2018) on Musk's relationship with OpenAI: Zilis's note that \"**They seem pretty set on doing the OpenAI thing. Elon can support or he can choose not to. … Greg and Ilya definitely don't want to do Tesla, Altman is increasingly that way, though could be convinced if Elon came to a plan they both believed in,**\" and Teller's \"**Understand your point about positive vs negative but still just feels like bs when they say the reason they don't want to join Tesla is he hasn't done his homework. I haven't heard them articulate a compelling vision either tbh.**\"\n\n## Document type\n**Email thread, plain text, multiple messages** between Sam Teller (Tesla chief of staff) and Shivon Zilis (then at OpenAI). Subject: \"Re: A few things.\" Bates OPENAI_MUSK00004332–00004333. Marked CONFIDENTIAL. Per the wiki reference index: \"DX 756 — limiting-instruction exhibit\" — i.e., admitted with a Federal Rules of Evidence limiting instruction (likely 105) restricting the jury's permissible use.\n\n## Logistics\n- **Trial admission:** Day 8 (May 6, 2026) — Murati live; Zilis live; Toner video begins. Used during Zilis's live testimony (the Day 8 Zilis sequence is in [[Key Themes]] §\"Day 8 — what Zilis added (and what the defense wanted from her)\").\n- **Box upload:** 2026-05-06 14:56:55 PT — Day 8 mid-afternoon Defense batch (clustered with [[DX 1048]]).\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~225 KB, 2 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `0756.pdf`.\n\n## Transcribed text\n\n> **From:** Sam Teller <steller@spacex.com>\n> **To:** Shivon Zilis <shivon@openai.com>\n> **Subject:** Re: A few things\n> **Date:** 2/9/2018 6:34:23 PM\n>\n> I know not literally homework, but I feel like they are only one degree less abstract than E. Maybe I'm wrong. If they say they have tons of clarity they are lying because no one knows! I get that E just need to have and communicate a little more of a positive vision. You are totally right about that.\n\n> **From:** Shivon Zilis <shivon@openai.com>\n> **To:** Sam Teller <steller@spacex.com>\n> **Subject:** Re: A few things\n> **Date:** Feb 9, 2018, at 10:31 AM\n>\n> Homework is not meant as literally as you're taking it. It's just that he talks in such vibrant terms about how, for example, Neuralink could evolve and doesn't currently have any directional plan for what to do in AI. The \"project centric\" thing in the doc I'm sending you will include thoughts on that. Assume SpaceX email for that? It's long so perhaps just look tomorrow when you have free mental space.\n\n> **From:** Sam Teller <steller@spacex.com>\n> **To:** Shivon Zilis <shivon@openai.com>\n> **Subject:** Re: A few things\n> **Date:** Feb 9, 2018, at 12:15 AM\n>\n> Yes re: focus\n>\n> I want to read doc, understand homework, etc\n>\n> Understand your point about positive vs negative but still just feels like bs when they say the reason they don't want to join Tesla is he hasn't done his homework. I haven't heard them articulate a compelling vision either tbh\n\n> **From:** Shivon Zilis <shivon@openai.com>\n> **To:** Sam Teller <steller@spacex.com>\n> **Subject:** Re: A few things\n> **Date:** Feb 8, 2018, at 11:54 PM\n>\n> They seem pretty set on doing the OpenAI thing. Elon can support or he can choose not to. They of course hope to be friends. Greg and Ilya definitely don't want to do Tesla, Altman is increasingly that way, though could be convinced if Elon came to a plan they both believed in. They all think Elon is an incredible human being but that he really hasn't done his homework AI / AGI and that really concerns them about working with him.\n>\n> Only outstanding item is structure of financing which Altman will know in a week. He said he wanted to run final plan by you and me, then to Elon. **It will either be private SAFT or a more traditional equity offering as a for profit or B Corp. Altman presumes he is still out but wants him to be aware at least.**\n>\n> For Elon, I will send you a document with all of the ways the world could potential unfold that have any merit. Some of them are pretty out there and difficult to achieve, but the point is they are conversation starters since Elon needs help making his mind think about this. I would be curious as to your reactions.\n>\n> Highest level bit: I can't see a viable path forward if Elon is not able to find a version of this he finds truly inspiring, meaning he has to think he is trying to create a good thing vs a .02% chance of a slightly less shitty outcome. Hard to take a viable shot at this while having the severely negative mental and emotional reaction. It will wreck him and not lead to good results. I think there could be versions of this that he finds inspiring, whether now or in time, but that's the biggest thing I'm trying to figure out for myself as well, as I've found this problem difficult to solve for myself too!\n>\n> Re: other things, I really want Tesla music to be awesome but I'm admittedly not the ideal person to work on that. There are so many things that would be happier than the AGI stuff but it feels important and we don't have anyone else to think about it!!! Feels correct to focus on it at least until Elon makes a decision, don't you think?\n\n> **From:** Sam Teller <steller@spacex.com>\n> **To:** Shivon Zilis <shivon@openai.com>\n> **Subject:** Re: A few things\n> **Date:** Feb 8, 2018, at 6:57 PM\n>\n> Thanks!\n>\n> On AI, do you have any more intel from Greg and Ilya? Are we mainly still trying to get Elon to think about and act on a plan? Recruit Altman? Read stuff from you?\n>\n> I think the emails are still good for Neuralink updates. Will be back in the rhythm next week! All hands with Q&A sounds good. Also let's block time next week to go through his old SpaceX and Tesla emails to extract great ones to share with Neuralink.\n>\n> Tesla I think ok for now. I'm close to the Jon stuff. Omead deep in production. Need you to stay close to Autopilot, which has been very helpful.\n>\n> Anything else you want to be doing that would make you happy?\n\n> **From:** Shivon Zilis <shivon@openai.com>\n> **To:** Sam Teller <steller@spacex.com>\n> **Subject:** A few things\n> **Date:** Feb 8, 2018, at 12:19 AM\n>\n> 1. I am guessing tomorrow not a great day for you to think about AI. Happy and ready to talk about it whenever you are but I understand that it's not top of mind — so queue me when you're ready! Everyone sort of in a holding pattern so not urgent, but next 2-5 days probably ideal.\n>\n> 2. If you can think of better ways to keep Elon up to speed about Neuralink let me know. I feel like the context emails used to be the best strategy but don't think that's working as well as it used to. Would love to keep him feeling like he is engaged and in the loop so if you have thoughts I'm all ears.\n>\n> Really think he should do an all hands soon so it's more front line than just me / you / Max giving him high level.\n>\n> 3. If there is Tesla stuff that needs handling given Jon and such let me know. I have been pretty heads down trying to think through AI options / help manage problems that pop up and provide direction while you guys have been away — but please refocus me at will!\n\n## Commentary\n\nDX 756 is admitted with a **limiting instruction** (per the wiki reference index) — meaning the jury was told to consider it only for a specific permitted purpose (likely state-of-mind or course-of-conduct rather than for the truth of the matter asserted). The substantive content cuts hard for defense's [[Key Themes]] §\"Unequivocal control\" / \"tale of two Elons\" theme: by **February 8, 2018**, Zilis (Musk's information conduit into OpenAI) was reporting to Teller (Musk's Tesla chief of staff) that **the founders had decided to do \"the OpenAI thing\"** — i.e., a separate raise either as a SAFT, a for-profit, or a B Corp — and that **\"Altman presumes he is still out but wants him to be aware at least.\"** Two things follow. First, by Feb 2018, Musk was already external to the founders' planning and the founders considered him out. Second, Zilis frames Musk's options as \"support or choose not to\" — there is no language of *deception*, only of *parallel paths*, which is corroborated by the Feb 2018 Karpathy \"**bury this in Tesla for stealth advantage**\" / DX 758 scenario set covered on Day 8 (see [[Key Themes]] §\"Bury this in Tesla for stealth advantage\"). Plaintiffs' answer is that the contemporaneous *internal* OpenAI record ([[Brockman Journal]]) tells a different story than what Zilis was reporting back to Teller.\n\n---\n*See also:* [[Day 8|Day 8 digest]] · [[Brockman Journal]] · [[Shivon Zilis]] · [[Sam Teller]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-757", "exhibit": "DX 757", "party": "Defense", "type": "Email thread", "admitted_trial_day": "Day 8 (May 6, 2026)", "uploaded_box_pt": "2026-05-06T14:56:55", "uploader": "Morrison Foerster", "pages": 1, "size_bytes": 367796, "source_pdf": "DX-757.pdf", "pdf_url": "https://media.mts-in.com/DX-757.pdf", "body_markdown": "# DX 757 — Feb 11, 2018 Musk → Zilis, \"Re: OpenAI and this upcoming week\" (\"I should talk to them this week\")\n\n> A one-page email exchange in which Shivon Zilis updates Musk that \"**Greg, Ilya, and Altman seem to be leaning towards continuing with OpenAI**\" — that they will choose between staying nonprofit, an equity fundraise, or a private token offering — and Musk replies \"**I should talk to them this week.**\"\n\n## Document type\n**Email thread, plain text, two messages.** Musk's \"I should talk to them this week\" reply on top of Zilis's update from Sun, 11 Feb 2018 11:17 PM. From: Elon Musk <erm@spacex.com>; to: Shivon Zilis <Shivon@spacex.com>; cc: Sam Teller <steller@spacex.com>; subject \"Re: OpenAI and this upcoming week\"; date Sun, 11 Feb 2018 23:19:22 -0800. Bates 2024MUSK-0004497.\n\n## Logistics\n- **Trial admission:** Day 8 (May 6, 2026 — Murati live; Zilis live; Toner video begins). No prior wiki citation; admission day inferred from Box upload and Zilis's live testimony.\n- **Box upload:** 2026-05-06 14:56:55 PT — Day 8 mid-afternoon defense batch (clustered with DX 619, DX 664, DX 724, DX 824, DX 900).\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~359 KB, 1 page.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `0757.pdf`.\n\n## Transcribed text\n\n> **From:** Elon Musk <erm@spacex.com>\n> **To:** Shivon Zilis <Shivon@spacex.com>\n> **Cc:** Sam Teller <steller@spacex.com>\n> **Subject:** Re: OpenAI and this upcoming week\n> **Date:** Sun, 11 Feb 2018 23:19:22 -0800\n>\n> I should talk to them this week\n\n> **From:** Shivon Zilis <shivon@spacex.com>\n> **To:** Elon Musk <erm@spacex.com>\n> **Cc:** Sam Teller <steller@spacex.com>\n> **Subject:** OpenAI and this upcoming week\n> **Date:** Feb 11, 2018, at 11:17 PM\n>\n> Elon,\n>\n> OpenAI hasn't fully decided anything but quick update on their thinking:\n>\n> Greg, Ilya, and Altman seem to be leaning towards continuing with OpenAI. They had a chance to speak with Andrej, per your recommendation, but they still seem pretty tied to OpenAI — especially Greg and Ilya.\n>\n> If they do stick with the current structure they will try to significantly ramp up fundraising. They're still not sure whether they'll stay a non-profit and focus on donations, figure out a structure for an equity fundraise, or do a private version of the token offering you guys had discussed before. They said they'd have more clarity in a week or two.\n>\n> Wanted to check in and see whether you would like to speak to any / all of them this week or whether you'd prefer to let them figure things out before they update you?\n\n## Commentary\n\nDX 757 is the **February 2018 inflection-point** email — the moment **after** the Sept 2017 control-fight ([[PX 157]], [[PX 151]]) and the [[PX 164|Jan 31 \"moral high ground\" email]], when Brockman, Sutskever, and Altman were still actively choosing between three structural options: (i) **stay nonprofit and focus on donations**, (ii) **equity fundraise** (the eventual 2019 OpenAI LP capped-profit), or (iii) **a private version of the token offering** (the ICO Brockman had floated and Musk had vetoed in [[PX 164]]). Defense uses this to argue that the for-profit conversion was, in early 2018, **still an open architectural question that the founders were sharing transparently with Musk through Zilis** — i.e., there was no covert \"stole-a-charity\" track. The \"they still seem pretty tied to OpenAI — especially Greg and Ilya\" line corroborates Brockman's defense reading of the [[Brockman Journal|journal entries]]: the founders were committed to OpenAI continuing in some structurally-mission-protective form, with a fundraise (donations vs. equity vs. token) as the implementation question. Plaintiffs, of course, will use the same \"**equity fundraise**\" line as the moment Brockman tells Musk the for-profit conversion is on the table, which then sets up the [[Key Themes|\"unequivocal control\"]] / DX-757-then-DX-758 sequence Schubert ran with Zilis on Day 8.\n\n---\n*See also:* [[Day 8|Day 8 digest]] · [[Shivon Zilis]] · [[PX 164]] · [[PX 157]] · [[Brockman Journal]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-758", "exhibit": "DX 758", "party": "Defense", "type": "Email + memo", "admitted_trial_day": "Day 8 (May 6, 2026)", "uploaded_box_pt": "2026-05-06T14:56:58", "uploader": "Morrison Foerster", "pages": 5, "size_bytes": 725314, "source_pdf": "DX-758.pdf", "pdf_url": "https://media.mts-in.com/DX-758.pdf", "body_markdown": "# DX 758 — Feb 13, 2018 Zilis \"OpenAI / Possible AI scenarios\" memo to Musk\n\n> Shivon Zilis's February 13, 2018 cover email and attached \"Possible Scenarios\" memo to Elon Musk laying out **nine** strategic options for Musk's AGI play — including **#4 \"OpenAI Becomes a B Corp Subsidiary of Tesla,\"** **#5 \"Altman as anchor for TeslaAI,\"** and **#9 \"OpenAI / TeslaAI concrete partnership and division of labor\"** — Schubert's centerpiece exhibit on Day 8 for the [[Key Themes#Tesla / xAI as Musk's own AGI play (the parallel)|\"parallel AGI play\"]] frame.\n\n## Document type\n**Email + 4-page attached memo, 5 pages total.** Cover email from Shivon Zilis <Shivon@spacex.com> to Elon Musk <erm@spacex.com>, cc Sam Teller <steller@spacex.com>, sent Tue 13 Feb 2018 02:17:35 -0800, subject \"OpenAI / Possible AI scenarios,\" attaching `Possible_Scenarios.docx`. Production-stamped CONFIDENTIAL; Bates SPX-006748–006752. Document ID DX-0758.\n\n## Logistics\n- **Trial admission:** Day 8 (May 6, 2026) — Murati live; Zilis live; Toner video begins. Per the wiki TSV: \"**DX 758 — Feb 2018 Zilis OpenAI possible scenarios memo.**\" Used by Schubert (defense) on Zilis cross — see [[Key Themes]] \"**DX 758 — 'OpenAI possible scenarios':** scenario 4 — OpenAI becomes a B Corp subsidiary of Tesla; scenario 5 — Altman as anchor for Tesla AI; scenario 9 — OpenAI/Tesla concrete partner and division of labor.\" (`5/6/2026 Testimony @ ~12:43–12:52 PT`.)\n- **Box upload:** 2026-05-06 14:56:58 PT — Day 8 mid-afternoon batch (clustered with DX 643, DX 707, DX 728, DX 925).\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~708 KB, 5 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `0758.pdf`.\n\n## Transcribed text\n\n> **From:** Shivon Zilis <shivon@spacex.com>\n> **To:** Elon Musk <erm@spacex.com>\n> **Cc:** Sam Teller <steller@spacex.com>\n> **Subject:** OpenAI / Possible AI scenarios\n> **Date:** Tue, 13 Feb 2018 02:17:35 -0800\n>\n> Hi Elon,\n>\n> Realize the AGI stuff is a bear to think about. When I make myself think it through it's honestly with the heaviest of hearts, so I can only imagine how difficult it is for you.\n>\n> You've said you aren't sure if there is or isn't a worthwhile experiment to run for an AGI counterbalance at this point. I realize how short you are on time – so I suspended disbelief for a moment to jot down a set of possible paths forward (attached).\n>\n> Please know that they are not pitches, but thought starters. Some are a bit bonkers but, as you've said, it's kind of Hail Mary time so why not at least think them through. Figured even seeing some of these scenarios and saying \"hell no\" doesn't hurt to help you figure out what does / doesn't feel right to you.\n>\n> Feel free to disregard if this kind of thing isn't helpful, or if you don't feel like thinking about AI. The point was merely to try to take some of the weight off of your shoulders for getting to a decision.\n>\n> Shivon\n>\n> ---\n>\n> **Possible Scenarios**\n>\n> **1. TeslaAI: Ace-Centric Strategy**\n> *What:* Do this entirely at Tesla, where E hires 1-2 best in class \"ace\" researcher cornerstones to shore up AI talent across both research and engineering.\n> *Why:* E may want to make a full court press and do it quickly. He likely needs at least one cornerstone he can trust to do that.\n> *Pros:* Already has Andrej, who isn't the cornerstone but a big piece. E could be architect vs. having to be in the weeds. This would also allow someone to be 100% time on leading it.\n> *Cons:* List of people is **very** short. Jeff Dean (Google) and Oriol Vinyals (DeepMind) are top of mind. If you could get Shane Legg (DeepMind) it would be a coup. This strategy also leaves the vision in the hands of the ace, in large part. That could be ok if it aligns well, but unsure if it will since a lot of these guys are really enjoying living the rockstar life vs. worrying about AGI.\n> *1st step:* Meet with the handful of researchers that could potentially run this. Figure out what would make them join and what their fastest path to safe AGI plan would be.\n>\n> **2. TeslaAI: Project-Centric Strategy**\n> *What:* E defines 1-2 projects that he thinks aren't being done that would lead to a faster path to safe AGI than what DM is up to. No small task but I think it's possible.\n> *Why:* Would be a good opportunity for have E think deeply about what the best technical plan and experiments are to create a true counterbalance. … *Pros:* If E does come up with something, it will almost certainly be better than Demis' or OpenAI's current plans. … *Cons:* Would take a lot of E's time so may have to spin up post Model 3? … *1st step:* **please take this with 11 grains of salt** — E figures out what subset of projects he thinks are directionally correct. Perhaps mildly amusing, but my best guess for what this looks like him creating some version of the simulation that he thinks we are in. … smart interplay between agents and rich/well-constructed environments will be the safest of the fast paths to AGI -- and no one is doing anything particularly intelligent on the environment front.\n>\n> **3. TeslaAI: Research -> Applications Strategy (1st Race vs. 2nd Race theory)**\n> *What:* Convince a bunch of the multi-agent / generalization type researchers to co-locate with engineers that run experiments on how to make their research useful in the real world. Try to shore up the folks doing multi-agent work with massive salaries and equity upside.\n> *Why:* … the probable path to getting to AGI first would be winning the race to building and transitioning research algorithms to the real world, since they'd be worth trillions pretty quickly (this is the 1st race). These are well sub AGI, but arguably the person who gets there first will have a lead and resources to win the race to true AGI (2nd race). …\n> *Pros:* Fits in nicely with for-profit company structure. Aligns upside. Would slow Demis because you'd have to pull from his most valuable talent group.\n> *Cons:* If you even believe the assumptions above, this probably doesn't happen for 2-5 years. Could be hard to argue for at public company as a result, when there appears to be so much other low hanging AI fruit that a layman investor would want one to focus on. Also, could ignite an arms race if done too publicly.\n>\n> **4. OpenAI Becomes a B Corp Subsidiary of Tesla**\n> *What:* Convince OpenAI to be a B corp subsidiary. I have no idea if this is possible but… brainstorming ftw.\n> *Why:* They will be shielded from normal Tesla pressures and would have the flexibility to fundraise externally specifically for mission or get internal funding.\n> *Pros:* Keep the band together. Would alleviate some of their concerns around building AGI at a public company.\n> *Cons:* No clue if they are still willing to work with them or if the structure is feasible.\n> *1st step:* Figure out of structure is possible then float by OpenAI.\n>\n> **5. Altman as anchor for TeslaAI**\n> *What:* Do TeslaAI properly, but have Altman run it.\n> *Why:* He is the only person I know other than Demis, Greg, and Ilya who wants to spend a significant chunk of his life on safe AGI specifically and is capable of a large-scale experiment. If everything is left as is, he will be CEO of OpenAI anyhow, so his fate is likely to run a big AI experiment in either case.\n> *Pros:* Altman will try to aim big. He, of his own admission, has some skill gaps but the will is there and he is always willing to learn. … If OpenAI doesn't have Altman they may just decide TeslaAI is the better option too.\n> *Cons:* Trust continues to be an issue. Is there a way to solve that? Altman is also not a researcher / practitioner in the field so would have to be paired. Andrej a good start but he'd need more too.\n> *1st step:* Develop a compelling plan, both in terms of ideas and funding, and discuss it with Altman. Alternatively, hear out Altman's plan and see if it's a fit to do at Tesla.\n>\n> **6. If you can't beat 'em join 'em (or, really, influence them)**\n> *What:* Find a way to get Demis. Seriously. Maybe he comes to Tesla somehow or DeepMind is spun out. Perhaps convince him that tech backlash is about to hit Google hard (I hope it is… Google is terrifying) and that he's better off working with E.\n> *Why:* Demis really does fanboy hard and I don't think he's immoral… just amoral. If he hung around E perhaps it would force him to think about humanity more.\n> *1st step:* Suspend disbelief for a hot second and REALLY think about it, lol.\n>\n> **7. Funding and advisory**\n> *What:* E does not build something himself, nor aligns time with OpenAI, but funds the suite of things he thinks are most likely to have a positive impact. He will learn a bunch and retain the opportunity to do something when the time is right.\n> *Why:* If he can't find a way to positively frame this for himself, doing the AI thing is probably terrible for the world over all. … He will also retain a ton of goodwill in the ecosystem for if he ever wants to do something himself.\n> *Pros:* Can still be friends with OpenAI, who are likely to get 100's of millions unless they completely screw up. Can be Switzerland.\n> *Cons:* E is really the only person who can be a counterbalance to Demis so that would be out. This is certainly conceding on some level as well.\n>\n> **8. Stick with OpenAI**\n> *What:* Altman becomes CEO. They do an equity offering instead of a private ICO.\n> *Why:* They still hope for a world where this can happen and would welcome it. Would take a lot of the weight off E's mind but he'd have to believe they are on the right trajectory.\n> *Pros:* They already have kickass researchers. They are upskilling on the rest.\n> *Cons:* He doesn't do his own thing. Do we think OpenAI is on a counterbalance trajectory? Even if they don't do ICO there still may be reputational flak for for-profit equity thing.\n> *1st step:* Decide if OpenAI could conceivably be a counterbalance if Altman is CEO. Decide if he's comfortable with idea of Altman being CEO in general.\n>\n> **9. OpenAI / TeslaAI concrete partnership and division of labor**\n> *What:* Tesla explicitly collaborates on hardware, gets help on factory/robotic automation from OpenAI. A bunch of employees are 50/50 but no one is fully poached. Would need to find a way to align some upside – would likely be in the 50/50 arrangement.\n> *Why:* Tesla can do hardware. But it will be better with a tight collaboration. OpenAI will really struggle to do hardware by themselves. World is clearly better if we share on that front.\n> *Pros:* Stay friends. You get two experiments for the foreseeable future.\n> *Cons:* Falls apart if E does his own lab, so should only be done if no line of sight to that. **If OpenAI acquires Cerebras, the hardware collaboration may be a moot point.**\n> *1st step:* Discuss with OpenAI\n\n## Commentary\n\nDX 758 is the most explicit contemporaneous Tesla-AGI scenario document in the case — and Zilis's authorship gives it credibility plaintiffs cannot easily impeach. Five of the nine scenarios (#1, #2, #3, #5, #9) presuppose that **TeslaAI** is the destination, and #4 explicitly contemplates **OpenAI becoming a B corp subsidiary of Tesla** — the structural inverse of plaintiffs' \"stole a charity\" frame, in which the founders are alleged to have hijacked the nonprofit. The exhibit anchors the [[Key Themes#Tesla / xAI as Musk's own AGI play (the parallel)|parallel-AGI-play]] sequence next to [[DX-728]] (Teller's Dec 2017 \"build up Tesla\" note), Zilis's Feb 2018 [[DX 761]] \"**actively try to move three or four people from OpenAI to Tesla**\" line (`042926TT.txt:5232`), and ultimately [[DX-853]] (\"**zero percent**\" Dec 31, 2018) and DX 844 (\"attempt that through Tesla instead\"). On Zilis's redirect she walked the jury back through these scenarios as *thought-starters* (\"not pitches\"), supporting plaintiffs' point that \"**for-profit entities were created on paper. … no practical step was taken to convert**\" (`5/6/2026 Testimony @ ~14:13 PT`). The Cerebras line in scenario 9 (\"If OpenAI acquires Cerebras, the hardware collaboration may be a moot point\") is a small but useful breadcrumb for the [[Key Themes#Greg Brockman's contributions (\"zero\")|Cerebras conflict]] thread that runs through Brockman's Day 6 cross.\n\n---\n*See also:* [[Day 8|Day 8 digest]] · [[Shivon Zilis]] · [[DX-728]] · [[DX-707]] · [[DX-853]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-761", "exhibit": "DX 761", "party": "Defense", "type": "Text messages", "admitted_trial_day": "Day 3 (April 29, 2026)", "uploaded_box_pt": "2026-04-29T15:36:42", "uploader": "Morrison Foerster", "pages": 3, "size_bytes": 215523, "source_pdf": "DX-761.pdf", "pdf_url": "https://media.mts-in.com/DX-761.pdf", "body_markdown": "# DX-761 — Feb 16, 2018 Musk-Zilis texts: \"actively try to move three or four people from OpenAI to Tesla\"\n\n> Musk's text-message instruction to Shivon Zilis the day after his break with OpenAI — stay \"close and friendly,\" but \"actively try to move three or four people from OpenAI to Tesla\" — and the candid acknowledgment that \"There is little chance of OpenAI being a serious force if I focus on Tesla AI.\"\n\n## Document type\n**Text messages** (Apple Short Message Report extract). One conversation, 21 messages on 2/16/2018, two participants: ERM Old Number (Elon Musk) and Shivon Zilis. Bates ZILIS-0002324–0002326.\n\n## Logistics\n- **Trial admission:** Day 3 (April 29, 2026) — Musk direct concludes; cross begins. Used by Savitt on cross of Musk for the \"Tesla as Musk's parallel AGI play\" theme. Cited in [[Key Themes]] as the \"Feb 2018 Musk-Zilis move-3-or-4-people text.\"\n- **Box upload:** 2026-04-29 15:36:42 PT — Day-3 late-afternoon batch.\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~216 KB, 3 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `761.pdf`.\n\n## Transcribed text\n\n> **Short Message Report**\n> Conversations: 1 · Total Messages: 21 · Date Range: 2/16/2018\n> Participants: ERM Old Number (Elon Musk), Shivon Zilis\n>\n> **Shivon Zilis** [4:01 AM]: Do you prefer I stay close and friendly to OpenAI to keep info flowing or begin to disassociate? Trust game is about to get tricky so any guidance for how to do right by you is appreciated.\n>\n> **ERM Old Number (Musk)** [4:04 AM]: **Close and friendly, but we are going to actively try to move three or four people from OpenAI to Tesla. More than that will join over time, but we won't actively recruit them.**\n>\n> **Shivon Zilis** [4:05 AM]: Ok. And, yes, was on the calls\n>\n> **ERM Old Number (Musk)** [4:06 AM]: If we go after four people, who besides the two I spoke to today?\n>\n> **Shivon Zilis** [4:08 AM]: Depends what you want to do. Btw, Andrej actually really doesn't think highly of Woj and thinks zero probability of robotic path leading to AGI, just FYI. If software side there is a guy named Igor who is pioneering multi-agent stuff.\n>\n> **Shivon Zilis** [4:09 AM]: I actually think some probability you could get Ilya if you wanted him, but don't know if you do\n>\n> **Shivon Zilis** [4:09 AM]: He has been a very good spiritual leader\n>\n> **ERM Old Number (Musk)** [4:09 AM]: Ok, I should really talk to Andrej\n>\n> **ERM Old Number (Musk)** [4:09 AM]: Ilya is way too bought into OpenAI\n>\n> **Shivon Zilis** [4:10 AM]: After you hung up this morning, Altman and Greg were ok. Ilya was visibly devastated\n>\n> **ERM Old Number (Musk)** [4:10 AM]: Really?\n>\n> **Shivon Zilis** [4:10 AM]: But I hear you. Just info so you have it\n>\n> **Shivon Zilis** [4:10 AM]: Yes.\n>\n> **Shivon Zilis** [4:12 AM]: May I please just share one thing so I can put it out of my mind?\n>\n> **ERM Old Number (Musk)** [4:13 AM]: **Whether for personal or strategic reasons, Ilya is right to feel that way. There is little chance of OpenAI being a serious force if I focus on Tesla AI.**\n>\n> **ERM Old Number (Musk)** [4:13 AM]: Sure\n>\n> **Shivon Zilis** [4:15 AM]: I think there are a lot of no-brainers to explore, and I put some in that doc, but the think that keeps calling out to me is there is a very low probability of a good future if someone doesn't slow Demis down. Slowing him down is the only non-negotiable net good action I can see. You don't realize how much you have an ability to influence him directly or otherwise slow him down. I think you know I'm not a malicious person but in this case it feels fundamentally irresponsible to not find a way to slow or alter his path\n>\n> **ERM Old Number (Musk)** [4:19 AM]: Best to talk by phone about this later tonight\n>\n> **ERM Old Number (Musk)** [4:19 AM]: I doubt I could do so in a meaningful way\n>\n> **Shivon Zilis** [4:21 AM]: Ok, yes, that would be good. And, ultimately up to you of course, but I really think you can so would like to at least make the case. In any case, I will sleep better at night for having tried!\n>\n> **Shivon Zilis** [6:48 AM]: Will leave you be on the Demis stuff. I'm sure it's hard to think about and you have so much on your shoulders all the time that I always feel terrible pushing me. I just needed to say it once since it's been plaguing me. Always here to talk anything through when it helps you though! Have a wonderful night!\n\n## Commentary\n\nDX 761 is one of the defense's three or four most-quoted exhibits and the centerpiece of the [[Key Themes]] \"Tesla as Musk's parallel AGI play\" arc — Savitt confronted Musk with the **\"actively try to move three or four people from OpenAI to Tesla\"** line on [[Day 3|Day 3]] cross at `042926TT.txt:5232`. The texts also undercut Musk's \"I gave them everything and walked away clean\" framing: the same morning he hung up on the OpenAI founders he is talent-poaching, holding the option on Ilya (\"way too bought into OpenAI\"), and conceding that Ilya is \"right to feel that way\" about him. Musk's own contemporaneous self-assessment — **\"There is little chance of OpenAI being a serious force if I focus on Tesla AI\"** — anticipates the \"zero percent\" forecast in the Dec. 31, 2018 [[DX-853]] email and the November 2018 Newell letter ([[DX-844]], \"attempt that through Tesla instead\"). Zilis's \"slow Demis down\" plea is also the seed of plaintiffs' DeepMind-rivalry narrative.\n\n---\n\n*See also:* [[Day 3|Day 3 digest]] · [[Key Themes]] · [[Shivon Zilis]] · [[PX 157]] · [[DX-844]] · [[DX-853]]\n"}
{"exhibit_id": "DX-766", "exhibit": "DX 766", "party": "Defense", "type": "Text messages", "admitted_trial_day": "Day 8 (May 6, 2026)", "uploaded_box_pt": "2026-05-06T14:56:58", "uploader": "Morrison Foerster", "pages": 4, "size_bytes": 350207, "source_pdf": "DX-766.pdf", "pdf_url": "https://media.mts-in.com/DX-766.pdf", "body_markdown": "# DX 766 — Feb 17, 2018 Zilis/Teller texts (Tesla AI announce, \"Edison\" meme, OpenAI poach list)\n\n> A 30-message single-day text thread on February 17, 2018 between Sam Teller and Shivon Zilis — coordinating Tesla AI's announce post draft, swapping the **\"If you are driving a stolen Tesla / would it be called Edison?\"** Tesla-vs-OpenAI meme, and recording that Musk had asked Karpathy \"**to send a list of top OpenAI people to poach.**\"\n\n## Document type\n**Text messages, Cellebrite-style \"Short Message Report\" extraction.** Four pages: cover sheet (1 conversation `e38384659185b038f04c5a48d801631a`, 30 messages, 2 participants — Sam Teller, Shivon Zilis) plus three pages of message-by-message transcript with timestamps in GMT+00:00, including an inline \"**Edison**\" meme image (IMG_2515.jpeg, 108 KB). Bates ZILIS-0002599–2602. Document ID DX-0766.\n\n## Logistics\n- **Trial admission:** Day 8 (May 6, 2026) — used during Zilis's live testimony. Per [[Key Themes]] § \"Tesla / xAI as Musk's own AGI play,\" DX 766 is the centerpiece of defense's parallel-AGI-play theme.\n- **Box upload:** 2026-05-06 14:56:58 PT — Day 8 mid-afternoon batch (clustered with DX 710).\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~342 KB, 4 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `0766.pdf`.\n\n## Transcribed text\n\n> **Conversation:** e38384659185b038f04c5a48d801631a\n> **Date Range:** 2/17/2018\n> **Participants:** Sam Teller, Shivon Zilis\n>\n> **Messages in chronological order (GMT +00:00):**\n>\n> **SZ** (Shivon Zilis) — 2/17/2018, 3:10 AM\n> Hey! What is best process?\n>\n> **SZ** — 2/17/2018, 3:10 AM\n> Email Elon whole draft post with proposed edits? Can let him know that they agreed to all-hands too\n>\n> **ST** (Sam Teller) — 2/17/2018, 3:11 AM\n> Yes\n>\n> **ST** — 2/17/2018, 3:11 AM\n> Sounds good\n>\n> **SZ** — 2/17/2018, 3:11 AM\n> Btw they will really hate my edited version\n>\n> **ST** — 2/17/2018, 3:11 AM\n> Yeah\n>\n> **ST** — 2/17/2018, 3:11 AM\n> Shrug\n>\n> **ST** — 2/17/2018, 3:11 AM\n> We can find somewhere in the middle\n>\n> **SZ** — 2/17/2018, 3:11 AM\n> Should we let him react to theirs?\n>\n> **SZ** — 2/17/2018, 3:11 AM\n> Or share the redo as well?\n>\n> **ST** — 2/17/2018, 3:26 AM\n> There's and put your sentence in bracket as suggested edit\n>\n> **ST** — 2/17/2018, 3:26 AM\n> Theirs*\n>\n> **SZ** — 2/17/2018, 3:35 AM\n> Deal\n>\n> **SZ** — 2/17/2018, 3:36 AM\n> [image attachment IMG_2515.jpeg, 108 KB — meme: top panel Tesla Roadster captioned \"**IF YOU ARE DRIVING A STOLEN TESLA**\"; bottom panel Thomas Edison portrait captioned \"**WOULD IT BE CALLED EDISON?**\"]\n>\n> **ST** — 2/17/2018, 3:36 AM\n> Lol\n>\n> **SZ** — 2/17/2018, 7:42 PM\n> Did E reach out to DoTA team or more people at OpenAI yet?\n>\n> **ST** — 2/17/2018, 7:58 PM\n> No idea..\n>\n> **SZ** — 2/17/2018, 8:08 PM\n> Hm, ok\n>\n> **ST** — 2/17/2018, 8:10 PM\n> WhyV\n>\n> **ST** — 2/17/2018, 8:10 PM\n> ?\n>\n> **SZ** — 2/17/2018, 8:11 PM\n> Altman indicated he had. But I think he was just looking for info\n>\n> **SZ** — 2/17/2018, 8:11 PM\n> They know Elon looking to poach 4-5 because Scott and Woj both heard and shared that\n>\n> **SZ** — 2/17/2018, 8:12 PM\n> So I think he is trying to figure out who others are (assumes Elon already know… but E doesn't)\n>\n> **SZ** — 2/17/2018, 8:12 PM\n> *knows\n>\n> **SZ** — 2/17/2018, 8:18 PM\n> Btw — you advice when you have a moment? E asked Andrej to send a list of top OpenAI people to poach\n>\n> **SZ** — 2/17/2018, 8:23 PM\n> And e said for me to help him with it on the call. I am not 100% sure this info isn't all getting back to OpenAI but should I nudge Andrej if he hasn't sent it or let him do his thing?\n>\n> **ST** — 2/17/2018, 8:31 PM\n> Yeah\n>\n> **ST** — 2/17/2018, 8:31 PM\n> Should tell OpenAI it's not a secret\n>\n> **ST** — 2/17/2018, 8:31 PM\n> That a couple people may want to come work at Tesla\n>\n> **ST** — 2/17/2018, 8:31 PM\n> Shouldn't be a covert operation\n\n## Commentary\n\nDX 766 is the closest thing the defense has to a smoking gun for the Tesla-AI-as-Musk's-own-AGI-play theme. Three days before Musk formally resigned the OpenAI nonprofit board (February 20, 2018; see [[PX 233]]), Musk's two closest conduits — Sam Teller (Musk family office) and Shivon Zilis (then on both Neuralink and OpenAI's roster, by way of Tesla) — are openly discussing: (i) Karpathy's pending Tesla AI announce post, with Zilis editing it down (\"**they will really hate my edited version**\"); (ii) the \"**Edison**\" meme — visually framing Tesla as the \"stolen\" rebrand of Edison, i.e., a shot at the OpenAI/Altman side; and (iii) Musk's instruction to Karpathy to **\"send a list of top OpenAI people to poach,\"** with Zilis worried the info is \"all getting back to OpenAI\" and Teller pragmatically suggesting they \"tell OpenAI it's not a secret. That a couple people may want to come work at Tesla. Shouldn't be a covert operation.\" Defense uses this thread, alongside [[DX 748]] (Brockman's \"**working on a self-driving car is not vector aligned with working on AGI**\" reply to Musk), to argue that the same week Musk was disengaging from OpenAI, he was building Tesla AI from poaches off OpenAI's roster. Plaintiffs' rebuttal is that Tesla AI is self-driving, not AGI — but the [[Key Themes]] § \"Tesla / xAI\" entry tracks this as ongoing through xAI's 2023 founding. Zilis's \"I am not 100% sure this info isn't all getting back to OpenAI\" is also useful for impeachment because it dates Zilis's ambivalence about her own dual-loyalty role.\n\n---\n*See also:* [[Day 8|Day 8 digest]] · [[Shivon Zilis]] · [[Sam Teller]] · [[DX 748]] · [[PX 233]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-773", "exhibit": "DX 773", "party": "Defense", "type": "Email thread", "admitted_trial_day": "Day 3 (April 29, 2026)", "uploaded_box_pt": "2026-04-29T15:36:47", "uploader": "Morrison Foerster", "pages": 1, "size_bytes": 362726, "source_pdf": "DX-773.pdf", "pdf_url": "https://media.mts-in.com/DX-773.pdf", "body_markdown": "# DX 773 — June 8, 2017 Musk → Teller \"Fwd: Andrej Karpathy\" (\"The OpenAI guys are gonna want to kill me\")\n\n> Musk's June 8, 2017 forward to Sam Teller of his earlier-evening exchange with Tesla autopilot lead Jim Keller, in which Musk announces he just hired **Andrej Karpathy from OpenAI to be Director of Tesla Vision** and quips, \"**The OpenAI guys are gonna want to kill me, but it had to be done…**\"\n\n## Document type\n**Email thread, plain text, two messages.** Musk's June 8, 2017 11:14:59 PM (likely PT) forward to Sam Teller, on top of a Jim Keller (`jkeller@tesla.com`) reply at 9:12:17 PM PDT, on top of Musk's earlier 8:27 PM PDT message to Keller. Subject \"Fwd: Andrej Karpathy.\" Bates 2024MUSK-0011151. Document ID DX-0773.\n\n## Logistics\n- **Trial admission:** Day 3 (April 29, 2026). Pre-trial wiki reference: **\"DX 773 — June 2017 Musk-Keller (OpenAI guys want to kill me).\"** Used during plaintiffs' Musk cross to anchor the Tesla-AI / OpenAI-talent-poaching theme.\n- **Box upload:** 2026-04-29 15:36:47 PT — Day 3 late-afternoon defense batch.\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~354 KB, 1 page.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `773 .pdf` (note trailing space).\n\n## Transcribed text\n\n> **From:** Elon Musk <erm@tesla.com>\n> **Sent:** 6/8/2017 11:14:59 PM\n> **To:** Sam Teller [steller@tesla.com]\n> **Subject:** Fwd: Andrej Karpathy\n>\n> Begin forwarded message:\n>\n> **From:** Jim Keller <jkeller@tesla.com>\n> **Date:** June 8, 2017 at 9:12:17 PM PDT\n> **To:** Elon Musk <erm@tesla.com>\n> **Subject:** Re: Andrej Karpathy\n>\n> > This is good news. We need somebody great.\n> > You probably heard Minwoo and Berta quit today. That's also good.\n> > I'm still up for driving the rest of the software stack. I have some of the best people at Tesla and with some careful moving of people I can strength autopilot and still get my other software driver, bring up, radar and testing done.\n> >\n> > Let's talk tomorrow after you talk to Chris and see what the next best step is.\n> >\n> > Jim\n>\n> ---\n>\n> On Jun 8, 2017, at 8:27 PM, Elon Musk <erm@tesla.com> wrote:\n>\n> > Just talked to Andrej and he accepted joining as Director of Tesla Vision, so anyone working on neural net software would report to him. Probable start date in a week or two.\n> >\n> > Andrej is arguably the #2 guy in the world in computer vision after Ilya. **The OpenAI guys are gonna want to kill me, but it had to be done…**\n> >\n> > Thank goodness. Lmk if you want to talk tonight or tomorrow.\n> >\n> > Will let Chris Lattner know tonight at 11:30. Not sure if he will stay or go, but I won't be all that sad if the latter.\n\n## Commentary\n\nDX 773 is a one-page Day-3 defense exhibit with high punching weight. In June 2017, *while* still funding OpenAI as a 501(c)(3) and *while* sitting on its board, Musk personally recruited OpenAI's #2 vision researcher to Tesla and acknowledged in writing the conflict (\"**The OpenAI guys are gonna want to kill me**\"). Defense uses this as the launch point of the parallel-AGI-play timeline that culminates in [[DX 748|Brockman's Feb. 2018 \"moral high ground\" reply]] and [[DX 749|Musk's \"OpenAI to attach to Tesla as its cash cow\" endorsement]] eight months later. The closing line about Chris Lattner — \"I won't be all that sad if the latter\" — also cuts against any \"hands-off Tesla AI\" framing on Musk's part. Three subtler points get airtime on cross: (i) Karpathy moved to Tesla *before* the August–September 2017 control negotiation that produced [[PX 157]], suggesting Musk had already begun the Tesla-AI hedge; (ii) \"**Andrej is arguably the #2 guy in the world in computer vision after Ilya**\" — Musk himself ranks Sutskever as #1, complicating later \"the founders contributed nothing\" plaintiffs framing; and (iii) Musk emails this from `erm@tesla.com` while OpenAI matters were typically routed via `erm@spacex.com` — defense reads the email-domain choice as itself revealing.\n\n---\n*See also:* [[Day 3|Day 3 digest]] · [[Andrej Karpathy]] · [[Ilya Sutskever]] · [[DX 749]] · [[DX 748]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-806", "exhibit": "DX 806", "party": "Defendants", "type": "Email thread", "admitted_trial_day": "Day 3 (April 29, 2026)", "uploaded_box_pt": "2026-04-29T15:37:43", "uploader": "Morrison Foerster", "pages": 2, "size_bytes": 402625, "source_pdf": "DX-806.pdf", "pdf_url": "https://media.mts-in.com/DX-806.pdf", "body_markdown": "# DX 806 — Apr 23, 2018 Musk reply \"Ok by me\" on Zilis update describing OpenAI's capped-profit financing structure\n\n> Musk's two-word reply (\"Ok by me\") on top of Shivon Zilis's April 23, 2018 status report telling him that OpenAI is \"definitely not doing an ICO but rather **equity that has a fixed maximum return**\" — and that Sam and Greg have asked Zilis to be on an \"informal advisory board.\" The earlier March 25, 2018 Zilis email in the same thread describes Altman's \"**capped at 50x**\" early thinking on the structure that eventually became the OpenAI, L.P. capped-profit waterfall.\n\n## Document type\n**Email thread, plain text, three messages.** Musk's Apr 23, 2018 2:06 AM PDT one-line reply on top of Zilis's Apr 23, 2018 1:49 AM email and her earlier Mar 25, 2018 11:03 AM email. Subject: \"Re: AI updates.\" From: Elon Musk <erm@spacex.com> to Shivon Zilis <shivon@neuralink.com>; cc Sam Teller <steller@spacex.com>. Bates 2024MUSK-0005604–5605.\n\n## Logistics\n- **Trial admission:** Day 3 (April 29, 2026) — Musk direct concludes; cross begins. Per the chunk wiki note: \"**DX 806 — capped-profit term-sheet OK by me**.\" Defense's headline document for the proposition that Musk personally **approved** OpenAI's capped-profit equity structure in real time, six years before suing over it.\n- **Box upload:** 2026-04-29 15:37:43 PT — Day 3 mid-afternoon batch.\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~393 KB, 2 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `806 .pdf`.\n\n## Transcribed text\n\n> **From:** Elon Musk <erm@spacex.com>\n> **To:** Shivon Zilis <shivon@neuralink.com>\n> **Cc:** Sam Teller <steller@spacex.com>\n> **Subject:** Re: AI updates\n> **Date:** Mon, 23 Apr 2018 02:06:23 -0700\n>\n> **Ok by me**\n\n> **From:** Shivon Zilis <shivon@neuralink.com>\n> **To:** Elon Musk <erm@spacex.com>\n> **Cc:** Sam Teller <steller@spacex.com>\n> **Subject:** Re: AI updates\n> **Date:** Apr 23, 2018, at 1:49 AM\n>\n> Updated info per a conversation with Altman. You're tentatively set to speak with him on Tuesday.\n>\n> **Financing:**\n> -He confirmed again that they are definitely not doing an ICO but rather **equity that has a fixed maximum return**.\n> -Would be a rather unique subsidiary structure for the raise which he wants to walk you through.\n> -Wants to move within 4-6 week on first round (probably largely Reid money, potentially some corporates).\n>\n> **Tech:**\n> -Says Dota 5v5 looking better than anticipated.\n> -The sharp rise in Dota bot performance is apparently causing people internally to worry that the timeline to AGI is sooner than they'd thought before.\n> -Thinks they are on track to beat Montezuma's Revenge shortly.\n>\n> **Time allocation:**\n> -I've reallocated most of the hours I used to spend with OpenAI to Neuralink and Tesla. This naturally happened with you stepping off the board and related factors — but if you'd prefer I pull more hours back to OpenAI oversight please let me know.\n> -**Sam and Greg asked if I'd be on their informal advisory board** (just Gabe Newell so far), which seems fine and better than the formal board given potential conflicts? If that doesn't feel right let me know what you'd prefer.\n\n> **From:** Shivon Zilis <shivon@neuralink.com>\n> **To:** Elon Musk <erm@spacex.com>\n> **Cc:** Sam Teller <steller@spacex.com>\n> **Subject:** AI updates\n> **Date:** Mar 25, 2018, at 11:03 AM\n>\n> **OpenAI**\n>\n> *Fundraising:*\n> -No longer doing the ICO / \"instrument to purchase compute in advance\" type structure. **Altman is thinking through an instrument where the 4-5 large corporates who are interested can invest with a return capped at 50x if OpenAI does get to some semblance of money-making AGI.** They apparently seem willing just for access reasons. He wants to discuss with you in more detail.\n>\n> *Formal Board Resignation:*\n> -You're still technically on the board so need to send a quick one liner to Sam Altman saying something like \"With this email I hereby resign as a director of OpenAI, **effective Feb 20th 2018**\".\n>\n> *Future Board:*\n> -Altman said he is cool with me joining then having to step off if I become conflicted, but is concerned that others would consider it a burned bridge if I had to step off. I think best bet is not to join for now and be an ambiguous advisor but let me know if you feel differently. They have **Adam D'Angelo as the potential fifth to take your place**, which seems great?\n>\n> **TeslaAI**\n> Andrej has three candidates in pipeline, may have 1-2 come in to meet you on Tuesday. He will send you a briefing note about them. Also, he's working on starter language for a potential release that will be ready to discuss Tuesday. It will follow the \"**full-stack AI lab**\" angle we talked about but, if that doesn't feel right, please course correct... is tricky messaging.\n>\n> **Cerebras**\n> Chip should be available in August for them to test, and they plan to let others have remote access in September. The Cerebras guy also mentioned that a lot of their recent customer interest has been from companies upset about the Nvidia change in terms of service (the one that forces companies away from consumer grade GPUs to enterprise Pascals/ Voltas). Scott Gray and Ilya continue to spend a bunch of time with them.\n\n## Commentary\n\nDX 806 is the defense's cleanest \"**Musk personally approved the capped-profit structure**\" exhibit. **Defense's read:** in March 2018 Altman is already telling Zilis (Musk's intermediary) about a \"**return capped at 50x**\" equity instrument; one month later Zilis confirms to Musk that OpenAI is \"**definitely not doing an ICO but rather equity that has a fixed maximum return**\" — i.e., the precise structure that became the OpenAI, L.P. limited partnership in 2019 and that Wu walked the jury through on Day 7 (see [[PX 203]] and [[Key Themes]] § \"Day 7 — the Microsoft economics\"). Musk's response — \"Ok by me\" — is the receipt. The same email also documents (i) Musk's voluntary reduction of OpenAI hours in favor of **Neuralink and Tesla** (\"I've reallocated most of the hours I used to spend with OpenAI\"), and (ii) the **TeslaAI** recruitment work being conducted in parallel by Karpathy with the \"full-stack AI lab\" framing — exactly the dual-track behavior plaintiffs describe as Musk's parallel AGI play (see [[Key Themes]] § \"Tesla / xAI as Musk's own AGI play\"). **Plaintiffs' read:** \"Ok by me\" is approval of an *informal advisory* role, not of the corporate structure; the actual capped-profit conversion happened in 2019 with terms that were materially different from the \"50x\" figure Altman mentioned in March 2018. The \"potential fifth\" to take Zilis's board place is **Adam D'Angelo**, who would later be one of the November 2023 OpenAI directors who fired Sam Altman (see [[Helen Toner]] testimony in [[Day 8|Day 8 digest]]). **Cerebras** is the same company at the center of plaintiffs' Brockman conflict-of-interest theory (see [[Key Themes]] § \"Greg Brockman's contributions\" and [[Brockman Journal]]).\n\n---\n*See also:* [[Day 3|Day 3 digest]] · [[PX 203]] · [[DX 715]] · [[Shivon Zilis]] · [[Sam Altman]] · [[Adam D'Angelo]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-810", "exhibit": "DX 810", "party": "Defense (OpenAI/MS)", "type": "IRS Form 990", "admitted_trial_day": "Day 4 (April 30, 2026)", "uploaded_box_pt": "2026-04-30T15:13:14", "uploader": "Morrison Foerster", "pages": 33, "size_bytes": 3249468, "source_pdf": "DX-810.pdf", "pdf_url": "https://media.mts-in.com/DX-810.pdf", "body_markdown": "# DX 810 — Musk Foundation 2016 Form 990-PF (FY 7/1/2016–6/30/2017)\n\n> The Musk Foundation's IRS Form 990-PF for the tax year covering OpenAI's first full operating year — the public-record vehicle through which Musk-Foundation-sourced contributions to OpenAI were channeled (via Vanguard Charitable donor-advised fund), and the document defense uses to anchor [[Key Themes#The \"donor-advised funds\" / tax-deduction wrinkle|the donor-advised-fund / tax-deduction]] line of cross.\n\n## Document type\n**IRS Form 990-PF (Return of Private Foundation)**, plus Form 2220 underpayment-of-estimated-tax schedule, Schedule D (Form 1041), Form 8949, and Attachments 1–12. EIN 77-0587507. Filing entity: **MUSK FOUNDATION**, P.O. Box 10195 Dept 863, Palo Alto, CA 94303. Signed under penalty of perjury by Elon Musk, President, on 5/3/18; preparer Ronald F. Gong, CTC | myCFO LLC. Bates BMO_00001865–00001897.\n\n## Logistics\n- **Trial admission:** Day 4 (April 30, 2026) — Musk cross + redirect; Birchall direct + cross. Used in the \"[[Key Themes#The \"donor-advised funds\" / tax-deduction wrinkle|tax-deduction wrinkle]]\" sequence and in the Birchall-CFO-of-Musk-Foundation cross (`043026TT.txt:5007, 5413`).\n- **Box upload:** 2026-04-30 15:13:14 PT — Day 4 mid-afternoon batch.\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~3.10 MB, 33 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `810.pdf`.\n\n## Transcribed text\n\n> *(Form 990-PF is a structured return; the substantive figures and disclosures are reproduced below from the boxes and attachments. Preparer line-items, computation worksheets (Forms 2220, 8949, Schedule D), and the IRS extension-approval notice on page 14 are summarized at the end.)*\n\n> **Part I — Analysis of Revenue and Expenses (FY 7/1/2016–6/30/2017)**\n> - Interest on savings & temporary cash investments (Atch 1, Goldman Sachs / Morgan Stanley): **$167**\n> - Dividends and interest from securities (Atch 2, Morgan Stanley): **$76**\n> - Net gain from sale of assets not on line 10: **$11,039,556** (gross sales price $11,143,507)\n> - Capital gain net income (Part IV, line 2): **$11,039,556**\n> - Other income (Atch 3, Pfizer Inc securities): **$16**\n> - **Total revenue:** **$11,039,815**\n> - Compensation of officers, directors, trustees: **$0**\n> - Legal fees (Atch 4): $9,000 — all charitable purposes\n> - Accounting fees (Atch 5): $10,000 (book) / $5,000 net investment / $5,000 charitable\n> - Taxes (Atch 6 — Form 199 fee $10; RRF-1 fee $300; CA Sec'y of State $20; federal estimated tax $220,000): **$220,330**\n> - **Total operating and administrative expenses:** $239,330 (book) / $5,000 (net invest.) / $14,330 (charitable)\n> - **Contributions, gifts, grants paid:** **$47,798,322**\n> - **Total expenses and disbursements:** **$48,037,652**\n> - Excess of revenue over expenses and disbursements: **−$36,997,837**\n> - Net investment income: **$11,034,815**\n\n> **Part II — Balance Sheets**\n> - Beginning-of-year total assets (book): **$3,340,644**\n> - End-of-year total assets (book): **$3,715,330**; **Fair Market Value: $362,932,713**\n> - Investments — corporate stock (Atch 7, Goldman Sachs X143-0 / Morgan Stanley): ending book $2,660,570; ending FMV **$361,877,953**\n> - Total liabilities: **$0**\n> - Net assets (unrestricted analog, line 29): $3,715,330\n\n> **Part III — Analysis of Changes in Net Assets**\n> - Beginning fund balances: $3,340,644\n> - Plus Part I line 27a: −$36,997,837\n> - Plus other increases (Atch 8 — *Unrealized gain/(loss) of stock transfer*): **$37,372,523**\n> - Ending fund balances: **$3,715,330**\n\n> **Part IV — Capital Gains (Tesla Inc share sales)**\n> - 20,000 shrs Tesla Inc — acquired 7/2/2010, sold 2/8/2017; gross sales $5,151,582; cost basis $53,172; **gain $5,098,410**\n> - 19,100 shrs Tesla Inc — acquired 7/2/2010, sold 5/19/2017; gross sales $5,991,925; cost basis $50,779; **gain $5,941,146**\n> - **Total long-term capital gain: $11,039,556**\n\n> **Part V/VI — Excise tax on net investment income:** $220,696 (2% rate); estimated tax payments $223,100; refund line $2,404 (credited to 2017).\n\n> **Part VII-A — Statements Regarding Activities (selected):**\n> - Political activities: **No** to all of 1a–1d.\n> - Activities not previously reported: No. Governing-instrument changes: No.\n> - States to which the foundation reports: **CA**.\n> - **Line 12 — Did the foundation make a distribution to a donor advised fund over which the foundation or a disqualified person had advisory privileges?** **YES** (see Atch 9).\n> - Public inspection requirements complied with: Yes.\n> - Website: **WWW.MUSKFOUNDATION.ORG**. Books in care of Ronald F. Gong; located at 2200 Geng Road, Suite 100, Palo Alto, CA 94303. Phone (650) 210-5000.\n\n> **Part VII-B — Statements (Form 4720):** All \"No\" — no self-dealing, no excess business holdings, no jeopardizing investments, no taxable expenditures, no prohibited tax-shelter transactions, no premiums on personal benefit contracts.\n\n> **Part VIII — Officers/Directors/Trustees (Atch 10):**\n> - **ELON MUSK — President & Director — 1.00 hr/week — $0 compensation, $0 benefits, $0 expense allowance.**\n> - **KIMBAL MUSK — Secretary, Treasurer & Director — 1.00 hr/week — $0 / $0 / $0.**\n> - Five highest-paid employees: NONE. Five highest-paid contractors: NONE.\n\n> **Part IX-A — Direct Charitable Activities:** **N/A.**\n> **Part IX-B — Program-Related Investments:** NONE.\n\n> **Part X — Minimum Investment Return:**\n> - Avg monthly FMV of securities: $271,809,780\n> - Avg monthly cash balances: $314,240\n> - Total: $272,124,020 → minus 1.5% cash deemed held ($4,081,860) → Net value of noncharitable-use assets: **$268,042,160**\n> - Minimum investment return (5%): **$13,402,108**\n\n> **Part XI — Distributable amount (after $220,696 excise tax):** **$13,181,412**.\n> **Part XII — Qualifying distributions:** **$47,812,652** (all from Part I col. (d) line 26).\n> **Part XIII — Undistributed Income / Excess Distributions Carryover:**\n> - Excess distributions carryover from 2011: $211,859; 2012: $787,011; 2013: $1,290,643; 2014: $565,180; 2015: $0 — **Total carryover into 2016: $2,854,693**.\n> - 2016 qualifying distribution applied to 2016 distributable amount: $13,181,412.\n> - **Remaining amount distributed out of corpus: $34,631,240.**\n> - **Excess distributions carryover to 2017: $37,274,074.**\n\n> **Part XIV — Private Operating Foundations:** NOT APPLICABLE.\n\n> **Part XV — Supplementary Information:**\n> - Substantial-contributor managers: **ELON MUSK**.\n> - Foundation makes contributions only to preselected charitable organizations and does not accept unsolicited requests for funds.\n> - **Atch 11 — Grants and Contributions Paid During the Year:**\n>   - **YC.ORG** — 335 Pioneer Way, Mountain View, CA 94041 — PC — *For scientific research* — **$10,000,000**\n>   - **VANGUARD CHARITABLE** — P.O. Box 9509, Warwick, RI 02889-9509 — PC — *For public charity support* — **$37,798,322**\n>   - **TOTAL CONTRIBUTIONS PAID: $47,798,322**\n\n> **Part XVI-A — Analysis of Income-Producing Activities:** Interest exclusion code 14: $167; dividends excl. code 14: $76; capital gain excl. code 18: $11,039,556; other (Pfizer securities) related/exempt-function: $16. **Subtotal: $11,039,815.**\n\n> **Part XVII — Transfers to and Transactions with Noncharitable Exempt Organizations:** **No** to all.\n\n> **Signature page:** Signed by **ELON MUSK, President**. Preparer: **RONALD F. GONG, CTC | myCFO, LLC**, 2200 Geng Road Suite 100, Palo Alto, CA 94303; PTIN P00191544; Firm EIN 77-0522698. Date 5/3/18.\n\n> **Attachment 9 — Distribution to a Donor Advised Fund:**\n> > \"THE FOUNDATION TREATED DISTRIBUTIONS MADE TO VANGUARD CHARITABLE GIFT FUND ('THE FUND'), A DONOR ADVISED FUND, AS QUALIFYING DISTRIBUTIONS DURING THE YEAR ENDED JUNE 30, 2017. THE FOUNDATION AND/OR A DISQUALIFIED PERSON HAD ADVISORY PRIVILEGES OVER THE FUND. THE FOUNDATION DOES NOT HAVE LEGAL OWNERSHIP OF THE FUND ASSETS. GRANTS MADE FROM THE FUND ARE MADE ONLY UPON APPROVAL BY VANGUARD CHARITABLE GIFT FUND. DISTRIBUTIONS TO THE FUND WILL ACCOMPLISH THE PURPOSES DESCRIBED IN IRC SECTION 170(C)(2)(B).\"\n\n> **IRS CP 211A notice (page 14):** \"We approved your Form 8868, Application for Extension of Time To File an Exempt Organization Return\"; new due date **May 15, 2018**.\n\n> *(Pages 28–33 contain Form 2220 underpayment-of-estimated-tax computation worksheets — annualized-installment method, 4-quarter required installment of $135,958 in Q4 — and Form 8949 / Schedule D (Form 1041) Long-Term capital-gain detail re-stating the same two Tesla share lots ($5,151,582 / $5,991,925 proceeds; basis $53,172 / $50,779; gain $5,098,410 / $5,941,146; aggregate $11,039,556). These are computational schedules that duplicate the Part IV figures.)*\n\n## Commentary\n\nDX 810 is the public-record proof of the **Vanguard Charitable Gift Fund** plumbing that runs through plaintiffs' \"stole a charity\" theory and through defense's tax-deduction cross. The 990-PF shows the Musk Foundation routed **$37,798,322** of FY2016–17 disbursements to Vanguard Charitable as a \"qualifying distribution\" — and the Atch 9 disclosure expressly states \"THE FOUNDATION DOES NOT HAVE LEGAL OWNERSHIP OF THE FUND ASSETS. GRANTS MADE FROM THE FUND ARE MADE ONLY UPON APPROVAL BY VANGUARD CHARITABLE GIFT FUND.\" This is the structural fact behind the [[Key Themes#The \"donor-advised funds\" / tax-deduction wrinkle|standing wrinkle]]: once Musk transferred to a DAF, he could direct but not retrieve. The exhibit also tees up the [[Key Themes#The $1B / $38M gap|$38M / $1B gap]] arithmetic — the bulk of the $38,191,066 ultimately tracked to OpenAI in PX 112A flowed through this Vanguard pipe in 2016–17, after Musk Foundation funded itself by selling 39,100 Tesla shares acquired in July 2010 for ~$104K combined and sold in 2017 for ~$11.14M. The $0-compensation, 1-hour-per-week officer disclosure (Elon and Kimbal Musk) is also the foundation-officer line that Wilson confronted Birchall with on cross (`043026TT.txt:5007–5039`). See [[Day 4|Day 4 digest]] and [[Key Themes]].\n\n---\n*See also:* [[Day 4|Day 4 digest]] · [[Jared Birchall]] · [[Key Themes]] · [[Musk Foundation]]\n"}
{"exhibit_id": "DX-819", "exhibit": "DX 819", "party": "Defense", "type": "Email", "admitted_trial_day": "Day 3 (April 29, 2026)", "uploaded_box_pt": "2026-04-29T15:36:47", "uploader": "Morrison Foerster", "pages": 2, "size_bytes": 450150, "source_pdf": "DX-819.pdf", "pdf_url": "https://media.mts-in.com/DX-819.pdf", "body_markdown": "# DX 819 — July 5, 2018 Shivon Zilis → Musk, \"AI Updates\" — capped-100x return, autonomous-weapons letter, Puerto Rico AGI conference\n\n> Zilis's \"top three\" briefing to Musk on July 5, 2018: (i) OpenAI is **closing ~$500M** \"into an OpenAI subsidiary that is **capped at 100x return** and will only pay out if OpenAI reaches some semblance of AGI\"; (ii) Tegmark's autonomous-weapons letter v2 (Musk listed first, ahead of Demis); (iii) Tegmark's January 2019 Puerto Rico AGI conference. Plus general AI updates on DeepMind, NVIDIA, CMU/Apple, Dota2.\n\n## Document type\n**Email**, plain text, single message from \"Shivon@spacex.com\" → Elon Musk (cc Sam Teller), 5 Jul 2018 21:49:05 -0700, Subject \"AI Updates.\" Bates 2024MUSK-0009985–0009986. Marked Highly Confidential-AEO.\n\n## Logistics\n- **Trial admission:** Day 3 (April 29, 2026) — Musk cross. Used by defense to anchor the [[Key Themes]] §\"$1B / $38M gap\" / capped-profit theme: the **100x cap** that Musk was told about in mid-2018 is the structural antecedent of the structure plaintiffs now allege \"looted\" the charity.\n- **Box upload:** 2026-04-29 15:36:47 PT — Day 3 late-afternoon Defense batch (clustered with [[DX 844]]).\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~440 KB, 2 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `819 .pdf` (note trailing space in filename).\n\n## Transcribed text\n\n> **From:** Shivon Zilis <Shivon@spacex.com>\n> **To:** Elon Musk <erm@spacex.com>\n> **Cc:** Sam Teller <steller@spacex.com>\n> **Subject:** AI Updates\n> **Date:** Thu, 5 Jul 2018 21:49:05 -0700\n> **Importance:** Normal\n>\n> Would read the top 3 if you're able. The rest are general AI updates so you can disregard if your brain is full.\n>\n> **OpenAI funding:**\n> Trying to close ~$500M fairly soon. Would be corporates and high net worths. Instrument is not the ICO, but instead **an investment into an OpenAI subsidiary that is capped at 100x return and will only pay out if OpenAI reaches some semblance of AGI**. Not a typical structure, obviously, but apparently it has been used in things like oil exploration where it's all or nothing + massive reward. Altman will write you as soon as it's more concrete.\n>\n> **Autonomous Weapons Letter v2**\n> Found out through OpenAI that Tegmark is doing an updated push on the autonomous weapons letter that you'd previously signed. The v2 letter is similar but more direct in its ask and he seems to be including all the previous and new signatories, along with institutions this time (DeepMind currently listed first). The site lists an embargo date of July 18th.\n>\n> Your name is still listed first, followed by Demis, so wanted to double check if:\n> a) You're still comfortable being listed?\n> b) You would prefer the list to be alphabetized or somehow reordered?\n> (Last year you were generally supportive but annoyed that you became the entire face of it, so wanted to check)\n> https://futureoflife.org/lethal-autonomous-weapons-pledge/\n>\n> *Note: OpenAI declined to sign before and are not signing now. Tegmark recently went to the office to convince them but they disagree on whether this letter will be net helpful or harmful.*\n>\n> **Puerto Rico AGI conference**\n> Speaking of Tegmark, he's doing another conference in Jan 2019, this time back in Puerto Rico. This one will be 100% AGI focused. He emailed asking for a phone call to pick your brain on it. Can tee it up or defer depending on your preferences?\n>\n> ——General AI Updates——\n>\n> **Physics and AI**\n> Found out at a physics and AI dinner the other night that Google (likely X) may spin up a lab in this space. Still unclear if this will actually happen, but if it did this guy would be one of the founding physicists.\n>\n> **Dota2**\n> Planning to go to The International again this year. Are working closely with Valve on high profile 5v5 matchup. They are more confident that their 5v5 bots will win this year than they were about their 1v1 bot winning at this time last year.\n>\n> **NVIDIA AI research lab**\n> Have heard from many that they've amassed a bunch of really talented people in the last year, though the research direction seems almost entirely on new methods for applied technologies in GPU acceleration and graphics. Jensen (at CVPR) seemed most proud of their work in progressive GANs and said the next big problems they're trying to nail are inpainting and slo-mo.\n>\n> **CMU / Academia**\n> Friend who's the director of AI research at Apple and still a prof at CMU and mentioned the entire AI faculty of that school has been decimated recently. Even the mediocre professors are getting multi-million dollar deals, and a lot of the new offers are coming from finance firms like JPMC who are trying to catch up on the AI trend.\n>\n> **DeepMind**\n> Seems like they got a bit scooped with the unexpected Dota2 5v5 timing, but still made their voice heard with Quake III capture of the flag — with the rather self-righteous title of \"the emergence of complex cooperative agents\" like they were the first to the mark.\n>\n> Also, rumor has it that, on top of the folks that secretly converse on Twitter DM because they don't trust Demis not to spy on their email and gchat, a part of the inner group also meets in a London coffee shop without cell phones to have in person discussions away from him. Heard this from both Altman and another friend.\n\n## Commentary\n\nDX 819 is defense's contemporaneous \"**Musk knew**\" exhibit on the **capped-profit subsidiary structure**. As of **July 5, 2018** — six months before Musk's December 31, 2018 \"zero percent: not 1 percent\" email (DX 853) about OpenAI being \"relevant to DeepMind/Google\" (see [[Key Themes]] §\"Tesla / xAI as Musk's own AGI play\") — Musk had been told in writing that OpenAI was \"trying to close ~$500M\" via \"an investment into an OpenAI subsidiary that is **capped at 100x return** and will only pay out if OpenAI reaches some semblance of AGI.\" That is the LP / capped-profit structure plaintiffs now characterize as the looting of the charity. The \"**purple box**\" term sheet litigation on Day 4 (see [[Key Themes]] §\"The 2018 'purple box' term sheet\") works against Musk on this same point: he was told, and he didn't sue. The DeepMind/Demis material is colorful but secondary; the main point of the exhibit is the **timing of Musk's knowledge** of the capped-profit conversion architecture. Pair with [[DX 827]] (the Aug 31, 2018 OpenAI LP Summary Term Sheet).\n\n---\n*See also:* [[Day 3|Day 3 digest]] · [[DX 827]] · [[Shivon Zilis]] · [[Elon Musk]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-824", "exhibit": "DX 824", "party": "Defense", "type": "Text messages", "admitted_trial_day": "Day 8 (May 6, 2026)", "uploaded_box_pt": "2026-05-06T14:56:58", "uploader": "Morrison Foerster", "pages": 4, "size_bytes": 225247, "source_pdf": "DX-824.pdf", "pdf_url": "https://media.mts-in.com/DX-824.pdf", "body_markdown": "# DX 824 — Aug 27, 2018 Zilis ↔ Reyna Ortiz texts (OpenAI fundraise / Reid Hoffman)\n\n> A four-page text-thread extract from August 27, 2018 — Shivon Zilis ↔ Reyna Ortiz (Musk's executive assistant) coordinating an upcoming Reid Hoffman meeting in connection with the OpenAI fundraise, including Zilis's \"**Reid is 100M-250M of the potential 500M round**\" and \"**But I am scared of them closing a round without Elon's approval so I just want to say something to them.**\"\n\n## Document type\n**Text messages, Cellebrite-style \"Short Message Report\" extraction.** Four pages (cover sheet + ~3 pages of message-by-message transcript with timestamps in GMT +00:00). Conversation hash `0c7ed51c65eaf94be7d4560e235fca27`. 40 messages on 8/27/2018. Two participants: Reyna Ortiz, Shivon Zilis (handles redacted). Bates ZILIS-0000500–0000503. Page 4 is fully redacted (black overlay).\n\n## Logistics\n- **Trial admission:** Day 8 (May 6, 2026 — Murati live; Zilis live; Toner video begins). No prior wiki citation; admission day inferred from Box upload and Zilis's live testimony. Likely used in connection with the [[Key Themes|\"ironclad agreement\"]] / \"funding freeze\" thread — see Zilis's contemporaneous \"**OpenAI is likely to realize this week that there are $5,000,000 in Q3 funding is, albeit correctly, on hold**\" note from August 2017 (a year earlier).\n- **Box upload:** 2026-05-06 14:56:58 PT — Day 8 mid-afternoon defense batch.\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~220 KB, 4 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `0824.pdf`.\n\n## Transcribed text\n\n> **Date Range:** 8/27/2018 — **Total Messages:** 40 — **Participants:** 2 (Reyna Ortiz, Shivon Zilis)\n> [Times in GMT +00:00]\n>\n> **Shivon Zilis** — 4:55 PM — All good to send you guys normal scheduling email?\n> **Shivon Zilis** — 4:55 PM — Do you know if Reid meeting happened?\n> **Reyna Ortiz** — 4:55 PM — Yes please – I think we will have to cram everything into tues. dont think wed is an option anymore.\n> **Reyna Ortiz** — 4:56 PM — Thinking AP might have to be in fremont just flagging you, but no one else\n> **Reyna Ortiz** — 4:56 PM — reid hoffman?\n> **Shivon Zilis** — 5:02 PM — Yep!\n> **Reyna Ortiz** — 5:06 PM — No he was supposed to talk to him on the phone I thought?\n> **Shivon Zilis** — 5:09 PM — Yes Sam emailed Elissa I think but I dunno if it happened\n> **Reyna Ortiz** — 5:10 PM — He emailed us both about a call for maybe this week\n> **Reyna Ortiz** — 5:10 PM — What's it about though\n> **Reyna Ortiz** — 5:10 PM — E's schedule is just so messy I'm not sure\n> **Reyna Ortiz** — 5:10 PM — Lol\n> **Reyna Ortiz** — 5:10 PM — What priority level is\n> **Shivon Zilis** — 5:19 PM — Good q. Altman hasn't sent docs for Elon's review\n> **Shivon Zilis** — 5:19 PM — I presume he is waiting for Reid call\n> **Shivon Zilis** — 5:19 PM — But I am scared of them closing a round without Elon's approval so I just want to say something to them\n> **Reyna Ortiz** — 5:21 PM — What does this have to do with Reid\n> **Shivon Zilis** — 5:23 PM — They are directly related\n> **Shivon Zilis** — 5:24 PM — **Reid is 100M-250M of the potential 500M round**\n> **Shivon Zilis** — 5:24 PM — Elon said he wanted to talk to Reid as a next step\n> **Reyna Ortiz** — 5:30 PM — Oh see I didn't know that\n> **Reyna Ortiz** — 5:30 PM — copy\n> **Reyna Ortiz** — 5:30 PM — Will ask sam\n> **Shivon Zilis** — 5:30 PM — K. I texted Sam about it on weekend but never heard back\n> **Shivon Zilis** — 5:30 PM — I will email you all the things\n> **Reyna Ortiz** — 5:31 PM — kk\n> **Reyna Ortiz** — 5:31 PM — Thank you!!\n> **Reyna Ortiz** — 6:17 PM — Sam is OOO and stressed. Omead and I will let you know plan lol\n> **Shivon Zilis** — 6:18 PM — I could tell from that email\n> **Shivon Zilis** — 6:18 PM — Is he ok?\n> **Reyna Ortiz** — 6:19 PM — Yea he just needs to leave us to do the schedule and trust everyone is communicating and when he isn't onsite, he has trouble like letting things go if that makes sense\n> **Shivon Zilis** — 6:20 PM — Yep for sure. Will he be in Bay Area this week or no?\n> **Reyna Ortiz** — 6:50 PM — Sam - yes but nt with us tomorrow\n> **Reyna Ortiz** — 6:50 PM — not*\n>\n> [Page 4 fully redacted with black overlay.]\n\n## Commentary\n\nDX 824 is a contemporaneous, unguarded **August 2018** Zilis text record showing two things defense wants in evidence. **First**, that a year *after* the September 2017 final-straw ([[PX 157]]) — and contemporaneously with the [[Key Themes|\"purple box\" term sheet]] dated Aug 31, 2018 ([[DX-827|DX 827]]) — Zilis was actively coordinating an OpenAI fundraise on Musk's behalf, with **Reid Hoffman positioned as $100–250M of a potential $500M round**, signaling Musk's continued involvement and approval-rights in OpenAI's fundraise architecture even after he had stopped funding. **Second**, Zilis's \"**I am scared of them closing a round without Elon's approval**\" line cuts both directions: defense reads it as evidence that Musk's approval was a courtesy not a control — Zilis was free-floating worried, not exercising contractual authority — while plaintiffs read the same line as evidence of an actual veto right Musk possessed in 2018 that the founders later evaded. The \"Sam … OOO and stressed\" / \"trouble like letting things go\" Altman color is incidental but shows the inside-the-team tone. The black-overlay redacted page 4 is consistent with the defense having produced beyond what was needed to admit the substantive Reid-Hoffman-fundraise content. See [[Key Themes]] § \"Day 8 — what Zilis added.\"\n\n---\n*See also:* [[Day 8|Day 8 digest]] · [[Shivon Zilis]] · [[Reid Hoffman]] · [[Sam Altman]] · [[DX-827]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-827", "exhibit": "DX 827", "party": "Defense", "type": "Term sheet", "admitted_trial_day": "Day 4 (April 30, 2026)", "uploaded_box_pt": "2026-04-30T15:13:15", "uploader": "Morrison Foerster", "pages": 6, "size_bytes": 633345, "source_pdf": "DX-827.pdf", "pdf_url": "https://media.mts-in.com/DX-827.pdf", "body_markdown": "# DX 827 — Aug 31, 2018 Altman → Musk OpenAI LP \"purple-box\" term sheet\n\n> Altman's August 31, 2018 cover email to Musk attaching the **OpenAI LP Summary of Principal Terms** — the foundational capped-profit term sheet with the **\"IMPORTANT WARNING\"** purple box (\"**It would be wise to view any investment in OpenAI LP in the spirit of a donation**\") — and Altman's contemporaneous \"**my current thought is that I won't take any equity**\" disclosure.\n\n## Document type\n**Email + 5-page attached term sheet, six pages total.** Cover email from Sam Altman <shga@ycombinator.com> to Elon Musk <erm@spacex.com>, cc Sam Teller and Shivon Zilis, dated Fri, 31 Aug 2018 11:30:15 +0200, subject \"OpenAI term sheet,\" attaching `OpenAI_LP_-_Summary_of_Principal_Terms.pdf`. Bates SPX-001642 through SPX-001647. Document ID DX-0827.\n\n## Logistics\n- **Trial admission:** Day 4 (April 30, 2026) — Musk cross + redirect; Birchall direct + cross. Per [[Key Themes]]: \"DX 827 — Aug 2018 purple-box LP term sheet.\" Used by defense during Musk cross to establish that Musk personally received and was offered participation in the capped-profit structure, and again by Birchall for the \"infamous purple box\" line (\"Pretty plain vanilla for-profit structure\").\n- **Box upload:** 2026-04-30 15:13:15 PT — Day 4 mid-afternoon batch (clustered with DX 857, DX 1284).\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~633 KB, 6 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `0827.pdf`.\n\n## Transcribed text\n\n> **From:** Sam Altman <shga@ycombinator.com>\n> **To:** Elon Musk <erm@spacex.com>\n> **Cc:** Sam Teller <steller@spacex.com>, Shivon Zilis <shivon@openai.com>\n> **Subject:** OpenAI term sheet\n> **Date:** Fri, 31 Aug 2018 11:30:15 +0200\n> **Importance:** Normal\n> **Attachments:** OpenAI_LP_-_Summary_of_Principal_Terms.pdf\n>\n> Elon--\n>\n> Please see attached, look forward to feedback.\n>\n> Also, **my current thought is that I won't take any equity. I'm not doing this for the money anyway, and I like the idea of being completely unconflicted and just focused on the best outcome for the world.** If it appeared at some point we weren't going to build AGI but were going to build something valuable, then maybe I'd want equity then.\n>\n> Sam\n>\n> ---\n>\n> **OpenAI LP — Summary Term Sheet**\n>\n> **IMPORTANT WARNING [purple box]:**\n>\n> > **\\*\\*Investing in OpenAI LP (the Partnership) is a *high-risk investment*\\*\\***\n> > **\\*\\*Investors could lose their capital contribution and not see any return\\*\\***\n> > **\\*\\*It would be wise to view any investment in OpenAI LP in the spirit of a donation, with the understanding that it may be difficult to know what role money will play in a post-AGI world\\*\\***\n> >\n> > As described herein and in the Limited Partnership Agreement, the Partnership exists to advance OpenAI Inc's mission of ensuring that safe artificial general intelligence is developed and benefits all of humanity. The General Partner's duty to this mission and the principles advanced in the OpenAI Inc Charter take precedence over any obligation to generate a profit. The Partnership may never make a profit, and the General Partner is under no obligation to do so. The General Partner is free to re-invest any or all of the Partnership's cash flow into research and development activities and/or related expenses without any obligation to the Limited Partners. See Section 6.4 of the Limited Partnership Agreement for additional details.\n>\n> **Mission**\n> OpenAI's mission is to ensure that artificial general intelligence — by which we mean highly autonomous systems that outperform humans at most economically valuable work — (AGI), when developed, is safe and benefits all of humanity. We will attempt to directly build safe and beneficial AGI, but will also consider our mission fulfilled if our work aids others to achieve this outcome.\n>\n> **Corporate Structure**\n> OpenAI LP will be a for-profit Delaware Limited Partnership managed by its General Partner, a single-member Delaware LLC controlled by OpenAI, Inc. (the Nonprofit)'s Board of Directors. At all times, no more than a minority of the Nonprofit's Board of Directors will be holders of any economic interest in OpenAI LP or in the employee holdings entity (described below).\n>\n> **Funding / Revenue**\n> *Initial Capitalization:* OpenAI LP will initially be capitalized by a contribution of assets from the Nonprofit. The Nonprofit will get an interest equivalent to that of a Limited Partner in the initial raise that is consistent with the value of its capital contribution.\n>\n> *Limited Partner Interests:* We will sell preferred capped Limited Partner Interests redeemable (on a first in, first out basis) for a multiple of the purchase price (adjusted for inflation as described below). Redemption will commence either (a) if and when we successfully create AGI capable of generating the requisite returns and the General Partner, in its sole and absolute discretion, determines that redemptions will begin, or (b) at a mutually agreeable pre-AGI alternative.\n>\n> Specifically, the fundraising rounds will be as follows:\n> 1. Initial raise: $500M, 100x target redemption\n> 2. Employee pool: $100B capped target redemption\n>    - Actual redemption amount will be based on the number of employee LP Interests issued, which will be determined (subject only to the aggregate cap) by the General Partner in its sole and absolute discretion.\n>    - The General Partner may increase the target redemption of the employee pool in its sole and absolute discretion.\n> 3. Second raise: $10+B, 15x expected target redemption.\n> 4. There may be additional fundraising rounds and/or an additional employee pool, though the interests of employee and/or investor Limited Partners cannot be diluted absent approval of a majority-in-interest of the affected class of Limited Partner (employee/investor).\n>\n> Distributions will be made as follows: First, distributions will be made to initial raise investors in proportion to their respective capital contributions until each investor has received an amount equal to its aggregate capital contributions, with such distributions counting against the investors' target redemption amount; Next, 25% of all distributions will be made to initial raise investors and employees in proportion to their respective target redemption amounts, and 75% will go to second raise investors in proportion to their respective capital contributions until each investor has received an amount equal to its aggregate capital contributions. Subsequently, all future payments will be in proportion to each investor/employee's respective target redemption amount. Distributions will continue until all investors and employees have received their respective target redemption amounts; and any amounts in excess of the target redemption amounts shall be distributed to or for the benefit of the Nonprofit in its capacity as a Limited Partner.\n>\n> For the first 10 years after the initial raise, the inflation rate will be determined by reference to the U.S. Consumer Price Index or such alternative U.S. governmental index as determined by the General Partner with the consent of a majority-in-interest of the Limited Partners (determined by reference to target redemption amounts). Subsequently, starting in year 11, the target redemption amounts will increase annually at the greater of the inflation rate or a simple interest rate equal to the London Interbank Offered Rate plus 2%.\n>\n> Any earlier redemption date must be approved by a majority-in-interest of the investor Limited Partners, as determined by reference to their target redemption amounts. Prior to redemption, all items of profits and loss will be allocated among the Partners in proportion to their target redemption amounts.\n>\n> *Revenue:* At some future date, OpenAI LP may generate revenue, primarily in the form of interest, dividends, capital gains, and royalties. To the extent it seeks to commercialize technology in a way that will generate significant revenue subject to the Unrelated Business Income Tax, OpenAI LP will create taxable \"corporate blocker\" subsidiaries to commercialize such technology. Should we generate revenue (through OpenAI LP or its subsidiaries) that is not reserved for Research & Development activities and/or other related expenses, at the sole discretion of the General Partner, such surplus revenue will be distributed to our Limited Partners…\n>\n> *Tax Distributions:* OpenAI LP will make customary tax distributions to all Partners in respect of the net taxable income of OpenAI LP, if any. The amount and timing of any tax distribution will be determined in the discretion of the General Partner.\n>\n> **Fiduciary Duties**\n> OpenAI LP exists to advance our mission using the principles expounded in OpenAI's Charter. We will employ the Partnership's assets to help us achieve these goals, such as through funding research and development activities, investing in companies and assets that may further the mission, and other related activities. The General Partner shall have the sole right to determine the amount of the Partnership's assets that shall be spent on and reserved for the expenditures in advancement of the mission.\n>\n> Our duty to these principles and the advancement of our mission takes precedence over any obligation to generate a profit. We may never make a profit, and we are under no obligation to do so. We are free to re-invest any or all of our cash flow into research and development activities and/or related expenses without any obligation to the Limited Partners. The General Partner will have discretion to establish and maintain reserves for such expenditures, consistent with the mission, so long as such discretion is exercised reasonably.\n>\n> **The fiduciary duties of the Nonprofit Board of Directors flow *exclusively* to the Nonprofit, not to the Limited Partners.**\n>\n> **Employee Comp.**\n> In addition to their salary, employees will be granted profit interests in an employee holding vehicle (detailed below) in exchange for their services. Certain employees of the Nonprofit will also be granted profits interests in the target redemption amount, if any, held by Nonprofit. It is intended that the profits interests will not give rise to tax liability at the time of grant. Grants will vest over 6 years, with 0% in the first year, followed by 20% vesting each year for the next five years. Employee interests will not be transferrable until the employee has been at OpenAI for one year. Employees are limited to selling a maximum of 25% of their vested interests per year. Each employee recipient will make a timely Section 83(b) election with respect to their profits interests.\n>\n> **Employee Holding Company**\n> We will create a separate holding company (OpenAI Holdings, L.P.) to issue profit interests to employees of OpenAI LP. OpenAI LP will issue mirror profits interests to OpenAI Holdings, and, as a result, OpenAI Holdings will be a partner in OpenAI LP (along with the investors). The profits interests held by employees in OpenAI Holdings thus will represent an indirect economic interest in OpenAI LP. By receiving profits interests in OpenAI Holdings rather than OpenAI LP, it is intended that employees will be spared certain negative tax consequences of holding a partnership interest directly in a partnership in which they are also employees (such as quarterly estimated tax filings, self-employment taxes and no longer being eligible to participate in certain employee benefit plans). Employees will receive W-2s from OpenAI LP and otherwise will be treated as employees at that level for tax purposes. Employees will receive K-1s from OpenAI Holdings in connection with their profits interests and otherwise will be treated as partners at that level for tax purposes.\n>\n> **Limitations Re: Transfer of Interests**\n> Notwithstanding any of the preceding, the transferability of LP Interests will be subject to limitations including: (1) transfers will be limited to a period occurring once per year; (2) transfers can only be between and among Limited Partners absent prior approval by the General Partner; (3) transfers will be limited so as to avoid material burdens upon the partnership or the General Partner under securities and other applicable laws; and (4) transfers will be limited to avoid partnership tax liability under US federal \"publicly traded partnership\" and other similar laws.\n>\n> **Information Rights**\n> Each Limited Partner will be furnished with annual financial statements within 90 days following the end of each financial year. All scheduled payments to Limited Partners will be considered confidential, and Limited Partners are not entitled to information regarding the interests of other Partners, provided that the Limited Partners will be entitled to such information as they need to self-organize for purposes of exercising their voting and other rights under the governing Limited Partnership Agreement. Confidentiality will be subject to customary exclusions for required governmental reporting; for example, the Nonprofit will be required to disclose certain information regarding OpenAI LP on its IRS Form 990, which will be available to the public.\n>\n> **Strategic Investment Agreements**\n> OpenAI LP is empowered to enter into strategic investment agreements with companies that become Limited Partners subject only to the caveat that those agreements must not clearly subvert the redemption waterfall described above nor clearly detract from the advancement of our mission.\n\n## Commentary\n\nDX 827 is the most economically and structurally important defense exhibit in the chunk. It establishes four facts the defense relies on heavily: (i) the **purple-box \"IMPORTANT WARNING\"** explicitly told every prospective LP investor \"**It would be wise to view any investment in OpenAI LP in the spirit of a donation**\" and that the GP's mission duty \"**take[s] precedence over any obligation to generate a profit**\" — central to defense's \"no concealment, no breach\" theme; (ii) Altman personally disclosed to Musk in writing on the cover email that **\"my current thought is that I won't take any equity,\"** later corroborated by Altman's [[PX 296]] line \"**Fwiw I personally have no equity and never have**\"; (iii) Musk himself was the original recipient — i.e., Musk knew about the LP, the 100x cap, and the $500M initial raise / $10+B second raise on August 31, 2018, materially before any plausible breach claim could accrue — central to defense's statute-of-limitations theory; and (iv) the **fiduciary-duties block** (\"**The fiduciary duties of the Nonprofit Board of Directors flow *exclusively* to the Nonprofit, not to the Limited Partners**\") is the structural feature plaintiffs' theory of betrayal must explain away, since it formally subordinates investor returns to mission. The 100x / 15x cap waterfall and the \"corporate blocker\" UBIT structure are the predicates for the later [[Watershed MOU]] and the $250B target-redemption stack discussed in Wu's Day 7 30(b)(6) testimony. Cross-reference: [[PX 24]] (Charter, April 2018), [[DX 748]] (\"moral high ground\" / non-profit duty), [[DX 862]] (the March 2019 public-announcement draft based on this term sheet), and [[PX 233]] (Zilis preview of the \"fixed maximum return\" structure).\n\n---\n*See also:* [[Day 3|Day 3 digest]] · [[Sam Altman]] · [[PX 24]] · [[DX 748]] · [[DX 862]] · [[PX 233]] · [[PX 296]] · [[Watershed MOU]] · [[Statute of Limitations]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-828", "exhibit": "DX 828", "party": "Defense", "type": "Email thread", "admitted_trial_day": "Day 4 (April 30, 2026)", "uploaded_box_pt": "2026-04-30T15:13:15", "uploader": "Morrison Foerster", "pages": 1, "size_bytes": 111540, "source_pdf": "DX-828.pdf", "pdf_url": "https://media.mts-in.com/DX-828.pdf", "body_markdown": "# DX 828 — Aug 31, 2018 Birchall → Zilis \"plain vanilla for-profit structure\" (the OpenAI LP term sheet reaction)\n\n> Jared Birchall's same-day forward-and-react to Shivon Zilis after Sam Altman sent Musk the OpenAI LP Summary Term Sheet ([[PX 236]]): \"**Pretty plain vanilla for-profit structure. So kinda hard to push a narrative that doesn't involve investors being very focused on ROI.**\" Plus the pointed \"**Did he/would he offer E a board seat?**\"\n\n## Document type\n**Email thread, plain text, three messages stitched.** Birchall's August 31, 2018 8:05:33 PM (UTC) reply on top of Zilis's same-day 4:06 AM \"FYI\" forward, on top of Sam Altman's August 31, 2018 2:30:15 AM PDT cover email to Musk attaching the OpenAI LP Summary of Principal Terms. Subject \"Re: OpenAI term sheet.\" Bates ZILIS-0004162. Document ID DX-0828.\n\n## Logistics\n- **Trial admission:** Day 4 (April 30, 2026). Pre-trial wiki reference: **\"DX 828 — Birchall to Zilis (plain vanilla for-profit).\"** Per [[Key Themes]]: Birchall on the same term sheet, \"*Pretty plain vanilla for-profit structure. So kind of hard to push a narrative that doesn't involve investors being very focused on ROI*.\" `(043026TT.txt:6682)`. Used during plaintiffs' Birchall cross by Wilson.\n- **Box upload:** 2026-04-30 15:13:15 PT — Day 4 mid-afternoon defense batch.\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~109 KB, 1 page.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `828.pdf`.\n\n## Transcribed text\n\n> **From:** Jared Birchall <jared@excession.com>\n> **To:** Shivon Zilis <shivon@openai.com>\n> **Subject:** Re: OpenAI term sheet\n> **Date:** Fri 8/31/2018 8:05:33 PM (UTC)\n>\n> Pretty plain vanilla for-profit structure. So kinda hard to push a narrative that doesn't involve investors being very focused on ROI. I'm a super fan of capitalism and making tons of money doing great things, but not sure if this correlates with the \"noble cause for humanity, not doing it to make money\" narrative.\n>\n> Did he/would he offer E a board seat?\n\n> **From:** Shivon Zilis <shivon@openai.com>\n> **To:** Jared Birchall <jared@excession.com>\n> **Subject:** Fwd: OpenAI term sheet\n> **Date:** Fri, Aug 31, 2018 at 4:06 AM\n>\n> FYI\n>\n> Begin forwarded message:\n\n> **From:** Sam Altman <shga@ycombinator.com>\n> **To:** Elon Musk <erm@spacex.com>\n> **Cc:** Sam Teller <steller@spacex.com>, Shivon Zilis <shivon@openai.com>\n> **Subject:** OpenAI term sheet\n> **Date:** August 31, 2018 at 2:30:15 AM PDT\n>\n> Elon--\n>\n> Please see attached, look forward to feedback.\n>\n> Also, my current thought is that I won't take any equity. I'm not doing this for the money anyway, and I like the idea of being completely unconflicted and just focused on the best outcome for the world. If it appeared at some point we weren't going to build AGI but were going to build something valuable, then maybe I'd want equity then.\n>\n> Sam\n\n## Commentary\n\nDX 828 is a small, clean exhibit that does outsized work in the [[Key Themes#The 2018 \"purple box\" term sheet|\"purple box\"]] dispute. **Birchall's same-day private read** — \"**Pretty plain vanilla for-profit structure. So kinda hard to push a narrative that doesn't involve investors being very focused on ROI**\" — is the defense's strongest \"Musk knew\" exhibit. It is the Musk Family Office's CFO, contemporaneously, calling the OpenAI LP what plaintiffs now claim it was secretly hidden from Musk: a for-profit. Defense pairs this with [[PX 236]] (the term sheet itself) and [[DX 827]] (the parallel Birchall-Musk thread) to argue Musk had constructive notice of the LP structure in August 2018 — six and a half years before he sued. Birchall's *posturing/virtue-signalling* deposition reading and Savitt's \"**infamous purple box**\" cross reference this exhibit. Two ironies: (i) Birchall is a self-described \"**super fan of capitalism**\" telling Musk's circle that the OpenAI structure is just capitalism, while Musk's own pleadings frame the same structure as a charitable-trust violation; and (ii) Birchall asks \"**Did he/would he offer E a board seat?**\" — a question that, taken with [[DX 1156]] (Musk's Feb 2025 bid to *acquire* the OpenAI IP with Zuckerberg), defense uses to argue Musk's grievance was always about access and control, not mission. Cross-reference: PX 235 / [[PX 236]] term sheet; Altman's \"**I won't take any equity**\" cover line — defense reads it as Altman walking the talk.\n\n---\n*See also:* [[Day 4|Day 4 digest]] · [[Jared Birchall]] · [[Shivon Zilis]] · [[Sam Altman]] · [[PX 236]] · [[DX 827]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-830", "exhibit": "DX 830", "party": "Defendants", "type": "Email thread", "admitted_trial_day": "Day 8 (May 6, 2026)", "uploaded_box_pt": "2026-05-06T14:57:01", "uploader": "Morrison Foerster", "pages": 3, "size_bytes": 622592, "source_pdf": "DX-830.pdf", "pdf_url": "https://media.mts-in.com/DX-830.pdf", "body_markdown": "# DX 830 — Aug 31–Sep 1, 2018 Zilis-Birchall-Teller \"Re: AI decisions\" memo on the OpenAI LP term sheet: tripartite structure, $300B+ aggregate target, \"wise to view in the spirit of a donation\"\n\n> A three-page Zilis memo email — \"meant as a comprehensive doc to throw all critical things in one place\" — laying out the August 2018 OpenAI LP term sheet for Musk's decision: the tripartite Inc / LP / Holdings structure, $500M first-round at 100x cap, $300B+ aggregate theoretical value, \"fiduciary duties of the Nonprofit Board flow exclusively to the Nonprofit, not the limited partners,\" and three possible outcomes for Musk: \"Let it go entirely,\" \"Negotiate for a different structure,\" or \"Participate in the current structure.\"\n\n## Document type\n**Email thread, plain text, three messages.** Shivon Zilis's Sep 1, 2018 1:19 AM PDT reply on top of Jared Birchall's Sep 1, 2018 12:21 AM email on top of Sam Teller's Aug 31, 2018 11:33 PM email on top of Zilis's original Aug 31, 2018 11:15 PM email. Subject: \"Re: AI decisions.\" From: Shivon Zilis <Shivon@spacex.com> to Jared Birchall <jared@excession.com>; cc Sam Teller <steller@spacex.com>, Reyna Ortiz <reyna@spacex.com>. Bates SPX-003576–003578.\n\n## Logistics\n- **Trial admission:** Day 8 (May 6, 2026) — Murati live; Zilis live; Toner video begins. Used by defense in connection with Zilis's live testimony as the contemporaneous record of how Musk's inner circle (Zilis, Teller, Birchall) actually evaluated the LP term sheet eight months before its execution. Key paired exhibit with [[DX 715]] (Oct 2017 Tesla \"bury\" memo) and [[DX 758]] (the OpenAI possible scenarios deck).\n- **Box upload:** 2026-05-06 14:57:01 PT — Day 8 mid-afternoon batch.\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~608 KB, 3 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `0830.pdf`.\n\n## Transcribed text\n\n> **From:** Shivon Zilis <Shivon@spacex.com>\n> **To:** Jared Birchall <jared@excession.com>\n> **Cc:** Sam Teller <steller@spacex.com>, Reyna Ortiz <reyna@spacex.com>\n> **Subject:** Re: AI decisions\n> **Date:** Sat, 1 Sep 2018 01:19:33 -0700\n>\n> Thank you both for looking.\n>\n> And good point on donations! I keep forgetting the building cost.\n>\n> Altman said early next week fine for their timeline so we're all good.\n\n> **From:** Jared Birchall <jared@excession.com>\n> **To:** Shivon Zilis <shivon@spacex.com>\n> **Cc:** Sam Teller <steller@spacex.com>, Reyna Ortiz <reyna@spacex.com>\n> **Subject:** Re: AI decisions\n> **Date:** Sep 1, 2018, at 12:21 AM\n>\n> Thanks for this. I tend to agree with Sam's thought of maintaining a seat and converting what has already been given into equity. As an investor, E would at least be able to keep tabs on progress, as well as keep the door open for greater influence if desired.\n>\n> One correction to Basic Fact #2, Elon's last donation was in August. **He donates about $300k p/mo.** But true that his last multi million dollar donation was in 2017.\n>\n> I really question what weight Disclaimers #3 and #4 really carry. In both cases an investor can choose to think or act differently with no apparent consequence. It'll be interesting to see what E thinks.\n>\n> I'm digesting the rest, but with funding it appears that the non-profit donations roll into the $500MM first round 1:1.\n>\n> I'll be back…\n\n> **From:** Sam Teller <steller@spacex.com>\n> **To:** Shivon Zilis <shivon@spacex.com>\n> **Cc:** Jared Birchall <jared@excession.com>, Reyna Ortiz <reyna@spacex.com>\n> **Subject:** Re: AI decisions\n> **Date:** Fri, Aug 31, 2018 at 11:33 PM\n>\n> Thank you. Very helpful. Let's discuss with Elon next week. **My instinct says it's better for him to maintain a seat at the table, since even if Google is way ahead, OpenAI is (I believe) the clear #2 at this point.**\n\n> **From:** Shivon Zilis <shivon@spacex.com>\n> **To:** Jared Birchall <jared@excession.com>, Sam Teller <steller@spacex.com>\n> **Cc:** Reyna Ortiz <reyna@spacex.com>\n> **Subject:** AI decisions\n> **Date:** Aug 31, 2018, at 11:15 PM\n>\n> Hey guys,\n>\n> Realize there's a lot on the go, but I think it's critical to ensure Elon is able to come to the right decision for himself on OpenAI now. **Once they close $500M raise in this structure it is going to be a big deal and relatively irreversible.**\n>\n> This is meant as a comprehensive doc to throw all critical things in one place so we can decide how to help him, but I'll note that I don't have a lot of experience with financial documents so please correct any and all things.\n>\n> *Basic Facts*\n>\n> 1. Elon does not currently have structural control since resigning from the board (did not opt for a proxy or any other persisting governance).\n> 2. He is the largest sole donor to date unless something has recently changed, though his last donation was in 2017.\n> 3. Altman clearly has a deep desire to work with Elon and have him aligned, but says he realizes that may not be possible and respects Elon's choices. **Greg and Ilya do not share those desires** but seem cool with either scenario and are leaving it to Altman.\n> 4. Elon needs to decide a) if / how he wants to be affiliated b) what edits to financial structure he'd advocate for if he's involved c) how he wants his donations to date to be treated.\n> 5. A call with Reid seems like a good option in all cases since that's by far the more information gathering and verifying conversation he can have.\n>\n> *Critical Elements of the Term Sheet*\n>\n> **Disclaimers:**\n> 1. High risk\n> 2. Potential for complete loss of capital\n> 3. **\"Wise to view\" in the spirit of a donation**\n> 4. Money may be meaningless in an AGI world\n> 5. Ability to reinvest any and all cash flow into R&D\n>\n> **Tripartite Structure:**\n> 1. OpenAI Inc (non-profit)\n> 2. OpenAI LP (investment vehicle)\n> 3. OpenAI Holdings (employee equity vehicle)\n>\n> **Funding:**\n> 0. Initialization Capital Rolls over from non-profit ($?? at 100x target — this is confusing)\n> 1. Initial Round: **$500M @ 100x target ($50B, \"Group 1\")**\n> 2. Employee Pool: In holding company, worth $100B at target ($100B minimum, increasable at GP discretion)\n> 3. Second Raise: $10B+ @ 15x target ($150B, \"Group 2\")\n> 4. Possible subsequent raises at TBD\n>\n> **Roundup:** We're talking about a target floor of **$300B++(!!)** of aggregate theoretical value. I realize money is meaningless in AGI world, I only flag this to illustrate the potential gravity of how people will perceive this entity.\n>\n> **Order of Distributions:**\n> 1. 100% to \"Group 1\" until 1x return\n> 2. 25% to \"Group 1\" and Employees until 100x reached, 75% to \"Group 2\" until 1x\n> 3. 100% to \"Group 2 until 15x reached\n> 4. **Any excess back to non-profit**\n>\n> **Other:**\n> Timeline: Nothing in first 10 years except if GP advocates for earlier? (Jared would be curious to get your read on exactly what they are saying… I'm having trouble parsing yearly floating rates vs. actual target payback)\n> Revenue: GP can reinvest revenue or distribute surplus revenue to investors\n> Fiduciary Duty: \"**fiduciary duties of the Nonprofit Board of Directors flow exclusively to the Nonprofit, not the limited partners**\"\n> Strategic Investment: Can accept strategic investment if it doesn't \"clearly subvert the redemption waterfall\" nor clearly detract from the advancement of our mission.\n>\n> *Possible Outcomes:*\n>\n> **Let it go entirely**\n> If he chooses this I may recommend he ask for a name change away from \"OpenAI\". That's so linked to his name and he came up with it. Since this is a huge transition from their current structure a rebranding doesn't seem ludicrous.\n>\n> **Negotiate for a different structure**\n> There are several things about this structure that may not pass muster with Elon, but I'm curious if he's willing to engage with it. It is very weird, but will note that the goal here is to find a mechanism of getting the billions needed for building AGI in the least harmful way, since there don't seem to be any harm-free ways. Donations at that order of magnitude have proved impossible (Hoffman mentioned his frustration with Gates on that front today), traditional for profit doesn't seem right, ICO was insane… so??? Seems worth at least thinking on what that structure is if not this one? **Elon's answer may still be the only passable capital source is \"at Tesla\" in which case they'd have to agree to disagree.**\n>\n> **Participate in the current structure**\n> If Elon does participate it seems like he'd need to regain some element of control, since the economic element of this is rather meaningless practically if executed correctly and doesn't give investors any control rights. **The fundamental constraint here is they will not allow him to have 51% control, so is it worth him advocating for a seat at the table?**\n>\n> *Next Steps*\n>\n> 1. Have Elon glance at the term sheet.\n> 2. Talk to Reid on fairly short order.\n> 3. TBD based on Reid outcomes.\n>\n> ------You are welcome to stop reading now unless your name is Jared------\n>\n> *Trees for the forest (detailed questions that may or may not matter depending on if Elon wants to engage):*\n>\n> **Term Sheet**\n> 1. *Heavy* control lies with the GP. Almost everything in the document is at the sole discretion of the GP. Is the GP the OpenAI non-profit board? If so, what is that going to look like over time? You have to trust those people a ton to buy into this if what you cared about what the good of infinity in the face of infinite wealth.\n> 2. \"Surplus\" revenue can be returned to investors at the discretion of the GP, and that capital would not count towards the target amount. This is mildly scary because it may create a situation of consistent pressure on the GP to create and disburse revenue.\n> 3. What is this \"corporate blocker\" concept in the revenue section?\n> 4. Initialization Capital… this is money rolling over from the non-profit but what is that and how much? I thought they were roughly out of funds so it's a bit confusing. Also a bit confusing why the non-profit itself gets an interest?\n> 5. Employee interest… if people are being distributed $100B in potential value will they be able to prevent themselves from advocating to actually get it? I'm a bit confused about equity compensation of employees if the goal is it's never meaningful. Maybe just pay people a hell of a lot?\n> 6. Strategic investment… this cause feels extremely loosely written. May be worth a clarifying edit?\n> 7. Was a little confused by Altman's comment of not taking equity now, but will take equity if it turns out OpenAI is not building AGI but something else revenue generating.\n> 8. What are the rough investment amounts and list of participants? Are Altman and Greg putting in anything?\n> 9. Is \"wise to view in the spirit of a donation\" sufficient for Elon?\n\n## Commentary\n\nDX 830 is the most detailed contemporaneous record produced at trial of how **Musk's inner circle actually understood the OpenAI LP term sheet** in the eight months before it was signed. **Defense's read:** in late August 2018, Zilis (Musk's intermediary), Teller (Musk's chief of staff), and Birchall (Musk's family-office head and the [[Musk Foundation]] CFO) were briefed on every material element of the structure that ultimately became OpenAI, L.P. (see [[PX 203]]) — the tripartite Inc/LP/Holdings architecture, the $500M @ 100x first round, the waterfall, the 10-year-no-distributions feature, and the \"fiduciary duties flow exclusively to the Nonprofit, not the limited partners\" clause. Musk's three options were laid out plainly: let it go, negotiate, or participate. Teller's instinct was to **maintain a seat**; Birchall's was to **convert prior donations to equity**; Zilis's central observation was that the only real obstacle was \"**they will not allow him to have 51% control**.\" Six years later, Musk would sue for the structure his deputies were here describing in neutral, transactional terms. **Plaintiffs' read:** the same memo says the quiet part out loud — Zilis flags that the term sheet's \"wise to view in the spirit of a donation\" disclaimer is a fig leaf, that the GP-heavy control structure means trusting the OpenAI board \"a ton,\" that \"surplus revenue\" distribution creates pressure to monetize, and that \"Greg and Ilya do not share [Altman's] desires\" to keep Musk involved. The \"$300B++(!!)\" target floor — flagged in Zilis's own punctuation as alarming — is the figure that, by 2024, the for-profit's commercial valuation would actually approach (see [[Key Themes]] § \"Day 7 — the Microsoft economics\" and [[PX 203]]). Birchall's \"**$300k p/mo**\" datum is the Musk Foundation's monthly OpenAI giving rate as of August 2018 — a useful peg for [[PX 91]] (the Fidelity Charitable grant compendium) and [[Key Themes]] § \"donor-advised funds.\" The \"**at Tesla**\" alternative Zilis flags is the same Tesla-absorbs-OpenAI thread running through [[DX 715]], [[DX 753]], and [[DX 758]]. Reid is Reid Hoffman.\n\n---\n*See also:* [[Day 8|Day 8 digest]] · [[PX 203]] · [[DX 715]] · [[DX 753]] · [[DX 758]] · [[Shivon Zilis]] · [[Sam Teller]] · [[Jared Birchall]] · [[Sam Altman]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-835", "exhibit": "DX 835", "party": "Defense (OpenAI/MS)", "type": "Email thread", "admitted_trial_day": "Day 8 (May 6, 2026)", "uploaded_box_pt": "2026-05-06T14:56:58", "uploader": "Morrison Foerster", "pages": 1, "size_bytes": 1056781, "source_pdf": "DX-835.pdf", "pdf_url": "https://media.mts-in.com/DX-835.pdf", "body_markdown": "# DX 835 — Sept 9–11, 2018 Zilis \"Considerations for Deer Creek and Pioneer this week\" thread\n\n> Shivon Zilis's September 2018 weekly-scheduling email to Sam Teller, Omead Afshar, Reyna Ortiz, and Jehn Balajadia about Musk's calendar across **Tesla Autopilot, Neuralink, and OpenAI** — including the bullet that Musk had **\"decided to be supportive in spirit of OpenAI but not participate in the new instrument,\"** and that the OpenAI Reid Hoffman call was \"no longer a need to schedule.\"\n\n## Document type\n**Email thread, plain text, four messages on Sept 9–11, 2018**, on tesla.com addresses (Zilis was operating from `shivon@tesla.com` by this point) — Sam Teller (steller@tesla.com), Shivon Zilis, Omead Afshar, Reyna Ortiz, Jehn Balajadia. Subject \"Re: Considerations for Deer Creek and Pioneer this week.\" Bates TESLA_000003416. Production-stamped HIGHLY CONFIDENTIAL — AEO.\n\n## Logistics\n- **Trial admission:** Day 8 (May 6, 2026) — Murati live; Zilis live; Toner video begins. Used in the Zilis cross alongside [[DX-716]], [[DX-754]], and [[DX-758]] to track Musk's transition from OpenAI participant to Tesla-AI principal in 2018 — and specifically the **Sept 11, 2018** moment Musk decided not to participate in OpenAI LP's \"new instrument\" (the August 2018 OpenAI LP term sheet — the [[Key Themes#The 2018 \"purple box\" term sheet|\"infamous purple box\"]]).\n- **Box upload:** 2026-05-06 14:56:58 PT — Day 8 mid-afternoon batch (clustered with [[DX-758]], DX 877).\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~1.03 MB, 1 page.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `0835.pdf`.\n\n## Transcribed text\n\n> **From:** Sam Teller <steller@tesla.com>\n> **Sent:** 9/11/2018 11:18:25 PM\n> **To:** Shivon Zilis [shivon@tesla.com]\n> **Cc:** Omead Afshar [oafshar@tesla.com]; Reyna Ortiz [reyna@tesla.com]; Jehn Balajadia [jehn@tesla.com]\n> **Subject:** Re: Considerations for Deer Creek and Pioneer this week\n>\n> Sounds good.\n>\n> ---\n>\n> On Sep 11, 2018, at 11:12 PM, Shivon Zilis <shivon@tesla.com> wrote:\n>\n> Ok. Spoke to Max. No need for any Neuralink tomorrow. Want me to give AP a heads up on Fremont?\n>\n> ---\n>\n> On Sep 11, 2018, at 11:02 PM, Sam Teller <steller@tesla.com> wrote:\n>\n> Not going to make it to DC or Neuralink tomorrow. Autopilot and V9 meetings can happen in Fremont. Anyone with important topics at Neuralink welcome to drive down.\n>\n> ---\n>\n> On Sep 9, 2018, at 10:23 PM, Shivon Zilis <shivon@tesla.com> wrote:\n>\n> Hi team!\n>\n> For this week…\n>\n> 1) Autopilot Weekly (1 hour)\n> - Deer Creek is of course always better for the team but if factory is what makes sense this week it's a less impactful week to do it. I also don't know if he has many other reasons to be in DC this week so if it's just AP we can have the team come over. If there is any control over when, the window of 12-3 is ideal for the team.\n>\n> 2) Neuralink (1.5 hours)\n> - Sam and I took a quick temperature on how Elon was thinking about Neuralink for the rest of the quarter and he mentioned he'd probably like to go this week. This should ideally be on Wednesday because there is an NHP surgery tomorrow and we should have outcome data by then. Tuesday, Friday, weekend are also fine if better — but Max is out Thursday so that's the only day that's suboptimal.\n>\n> Other notes:\n>\n> 1) AP / UI Airport Drive (push to next week)\n> - Checked in with Autopilot and this week isn't ideal for a drive from airport with Elon because the new UI isn't ready yet — so let's push that to when it is there to show him.\n>\n> 2) OpenAI Reid Hoffman call (no longer a need to schedule)\n> - Just for awareness, Elon does not need to do the Reid call because he's decided to be supportive in spirit of OpenAI but not participate in the new instrument.\n>\n> 3) AP safety (likely will be next week, some possibility it's 5-10 mins this week after AP meeting)\n> - Final review set for tomorrow at 1pm. If all ok with numbers across the team we can clear it this week. More likely that we'll need a few edits and opt to get final clearance from Elon next week or over email. Will report back on if team is in accord after tomorrow's meeting.\n\n## Commentary\n\nDX 835 is the contemporaneous internal note that pins Musk's decision **not to participate in the OpenAI LP \"new instrument\"** — the August 2018 capped-profit term sheet (see [[DX-827]] / the \"purple box\") — to a discrete week in **September 2018**, with the substitution that he would remain \"supportive in spirit.\" Defense uses this for two compounding points: (i) it places Musk's withdrawal from the for-profit conversion months before plaintiffs' theory of breach matures, contradicting the [[Key Themes#\"Captured by Microsoft\" (statute-of-limitations fight)|2020/2023 SOL story]]; and (ii) the email is sent from a `shivon@tesla.com` address, with Tesla Autopilot and Neuralink scheduling threaded into the same message — the operational picture of [[Key Themes#Tesla / xAI as Musk's own AGI play (the parallel)|\"actively try to move three or four people from OpenAI to Tesla\"]] from [[DX 761]] and the [[DX-758|Feb 2018 \"Possible Scenarios\"]] memo, now a quarter year into execution. Plaintiffs' read is the obverse: \"supportive in spirit\" is the precise phrase the OpenAI LP \"purple box\" itself used to characterize investments — and Musk's decision to abstain from the LP recapitalization is consistent with his contemporaneous belief he had funded a *nonprofit*, not a capped-profit vehicle. See [[Day 8|Day 8 digest]] and [[Key Themes]].\n\n---\n*See also:* [[Day 8|Day 8 digest]] · [[Shivon Zilis]] · [[DX-758]] · [[DX 761]] · [[DX-827]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-844", "exhibit": "DX 844", "party": "Defense", "type": "Email thread", "admitted_trial_day": "Day 3 (April 29, 2026)", "uploaded_box_pt": "2026-04-29T15:36:47", "uploader": "Morrison Foerster", "pages": 5, "size_bytes": 797412, "source_pdf": "DX-844.pdf", "pdf_url": "https://media.mts-in.com/DX-844.pdf", "body_markdown": "# DX 844 — Nov 2, 2018 Musk → Gabe Newell, \"I lost confidence that OpenAI could muster the resources … and decided to attempt that through Tesla instead\"\n\n> Musk's clean Nov 2, 2018 statement to Gabe Newell that his OpenAI involvement was \"**very limited at this point,**\" that he had \"**lost confidence that OpenAI could muster the resources to serve as an effective counterweight to Google/Deepmind**,\" and had \"**decided to attempt that through Tesla instead**\" — the cleanest single-sentence version of defense's \"Tesla / xAI as Musk's own AGI play\" theme.\n\n## Document type\n**Email thread, plain text, two messages.** Gabe Newell (Valve) → Musk, Oct 31, 2018 4:17 PM, \"Two things\"; Musk reply Nov 2, 2018 12:57:56 AM (\"Re: Two things\"); Newell reply Nov 2, 2018 6:40:29 PM. Bates 2024MUSK-0010577–0010581.\n\n## Logistics\n- **Trial admission:** Day 3 (April 29, 2026) — Musk direct concludes; cross begins. Per the wiki reference index: \"DX 844 — Nov 2018 Musk to Newell (attempt that through Tesla).\" See [[Key Themes]] §\"Tesla / xAI as Musk's own AGI play\" — this is the document Schubert quotes at `042926TT.txt:5170`.\n- **Box upload:** 2026-04-29 15:36:47 PT — Day 3 late-afternoon Defense batch (uploaded same time as [[DX 819]]).\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~779 KB, 5 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `844.pdf`.\n\n## Transcribed text\n\n> **From:** Elon Musk <erm@spacex.com>\n> **Sent:** Friday, November 2, 2018 12:58 AM\n> **To:** Gabe Newell\n> **Subject:** Re: Two things\n>\n> Sorry for the delay in responding. Sometimes i get a little backlogged on email. My cell is [redacted].\n>\n> Sure, it would be great to meet Hideo Kojima and he's welcome to see the rocket factory. No problem to send him my email.\n>\n> Best person to talk to at Neuralink is **Max Hodak (max@neuralink.com)**, who is the de facto head of day to day operations. **Shivon Zilis is also worth talking to (predictably at shivon@neuralink.com).**\n>\n> We've made some pretty insane technical progress. This is highly confidential, but we're now able to implant ~6000 electrodes in a monkey brain with decent signal/noise. Moreover, the electronics are compact enough to be flush with the skull and the only thing visible is the USB-C opening and slight surround. Very trippy. Just like Neuromancer.\n>\n> **Regarding OpenAI, my involvement is very limited at this point. I still provide some financial support and get verbal and email updates every few weeks from Sam Altman, but don't spend time there. I lost confidence that OpenAI could muster the resources to serve as an effective counterweight to Google/Deepmind and decided to attempt that through Tesla instead. We have cash flow on the order of billions of dollars per year to build hardware that hopefully has at least a dark horse chance to keep Google honest. Probably worth talking about at some point.**\n>\n> Elon\n\n> **From:** Gabe Newell\n> **To:** Elon Musk\n> **Sent:** Wednesday, October 31, 2018 4:17 PM\n> **Subject:** Two things\n>\n> 1) Hideo Kojima (*Metal Gear* series, a real visionary in our field) was here at Valve talking about his new game, and he mentioned the importance he places on future work in AI. I said I'd be happy to introduce him to the people at OpenAI and spoke enthusiastically about the work that team has done with us (I'm not sure how involved you are with them nowadays). The second thing is he was talking about how much he wants to go into space, and I offered to introduce him to you. He'd love to get a SpaceX tour.\n>\n> 2) For a long time I thought neuromodulation (e.g. rTMS) was weird, mainly because I had an unsophisticated understanding of a bunch of aspects of the brain. I've more or less done a 180, and think there is a significant near-term consumer market. Is this something I should bring up with the Neuralink team? If so, anyone in particular I should chat with there?\n>\n> Hope you are well.\n\n> **From:** Gabe Newell\n> **To:** Elon Musk\n> **Sent:** Friday, November 2, 2018 6:40:29 PM\n> **Subject:** RE: Two things\n>\n> Thanks. I've sent mail to Hideo, Max, and Shivon.\n>\n> Happy to talk about Tesla and AI when you're ready.\n\n## Commentary\n\nDX 844 is the cleanest single-sentence version of the [[Key Themes]] §\"Tesla / xAI as Musk's own AGI play\" / \"parallel AGI play\" theme: by **November 2, 2018** — six weeks before Musk's December 31, 2018 \"zero percent: not 1 percent\" email to the OpenAI founders (DX 853) — Musk had **already privately reached the conclusion** that OpenAI was insufficient and had \"**decided to attempt that through Tesla instead**.\" Plaintiffs use this exhibit two ways: (i) Musk's \"**I lost confidence**\" language is the same kind of subjective-state-of-mind framing that supports his \"phase one / phase two / phase three\" timeline (see [[Key Themes]] §\"Musk's three phases\"), so the date stamp helps fix when he started the transition to phase two; (ii) the fact that Musk was telling Gabe Newell that he had decided to attempt AGI through Tesla in November 2018, while still publicly framing OpenAI as on-mission and continuing to take tax deductions on contributions through the Musk Foundation (see [[PX 60]], [[DX 638]]), supports plaintiffs' \"tale of two Elons\" / inconsistent-position theme. Defense's response on cross is that nothing in DX 844 is inconsistent with the public Charter — Musk simply diversified his bets — and that this is exactly the conduct of someone who **walked away** from OpenAI rather than someone who was relying on OpenAI as a charitable trust beneficiary. Pair with [[DX 853]] (Dec 31, 2018, \"zero percent: not 1 percent\"), Musk's [[DX 1444]] tweet (\"Tesla will be one of the companies to make AGI\"), and the Day 8 Karpathy \"poach list\" + Zilis \"bury this in Tesla for stealth advantage\" sequence.\n\n---\n*See also:* [[Day 3|Day 3 digest]] · [[DX 819]] · [[DX 1444]] · [[Shivon Zilis]] · [[Elon Musk]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-849", "exhibit": "DX 849", "party": "Defense", "type": "Text messages", "admitted_trial_day": "Day 4 (April 30, 2026)", "uploaded_box_pt": "2026-04-30T15:13:15", "uploader": "Morrison Foerster", "pages": 2, "size_bytes": 129822, "source_pdf": "DX-849.pdf", "pdf_url": "https://media.mts-in.com/DX-849.pdf", "body_markdown": "# DX 849 — Nov 30, 2018 Birchall ↔ Greg Smithies (Neuralink) iMessage re: Pioneer Building rent (\"to pass nonprofit audits\")\n\n> A two-page iMessage transcript between Jared Birchall and Greg Smithies of Neuralink on November 30, 2018 — Birchall asking how the Pioneer Building rent reimbursement push is being driven and Smithies replying it is \"**openai accountants telling Chris he has to do it (so they can pass non-profit audits)**\" with a structural workaround: \"**we owe openai that rent, but then openai would technically owe that portion back to E**.\"\n\n## Document type\n**Text messages, Cellebrite-style \"Short Message Report\" extraction.** Two pages (cover sheet + one page of transcript). One conversation, 7 messages, 2 participants: Jared Birchall (custodian) and Greg Smithies. Date range 11/30/2018. Bates BIRCHALL-0000061–62.\n\n## Logistics\n- **Trial admission:** Day 4 (April 30, 2026 — Musk cross + redirect; Birchall direct + cross). Per [[Key Themes]]: \"DX 849 — Nov 2018 Birchall-Smithies email re: rent\" and \"Neuralink shared the building rent-free until OpenAI's auditors raised an issue (DX 849, 'to pass nonprofit audits')\" `(043026TT.txt:6360–6403)`. Quoted in [[Key Themes|wiki quotes]]: \"Sitting down this morning with E. **How much of the Pioneer stuff is Chris pushing for a repay versus you pushing for a fair/proper accounting?**\"\n- **Box upload:** 2026-04-30 15:13:15 PT — Day 4 mid-afternoon defense batch (clustered with DX 539, DX 1083, DX 1156, DX 1285, DX 516).\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~127 KB, 2 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `849.pdf`.\n\n## Transcribed text\n\n> **iMessage Chat:** [phone number redacted]\n> **Date Range:** 11/30/2018 — **Total Messages:** 7 — **Custodian:** Birchall, Jared — **Participants:** Jared Birchall, Greg Smithies\n> [Times in GMT +00:00]\n>\n> **Jared Birchall** — 11/30/2018, 3:58 PM\n> Sitting down this morning with E. How much of the pioneer stuff is Chris pushing for a re-pay vs you pushing for fair/proper Accounting? I guess where I'm going with this is, if he says there is no way we are paying for this, how big of a deal is that?\n>\n> **Greg Smithies** — 11/30/2018, 5:10 PM\n> Umm I'd say the main driver is **openai accountants telling Chris he has to do it (so they can pass non-profit audits)**. Because I didn't know if we wanted this one way or the other I've been pretty mum about trying to push the number or methodology in either direction\n>\n> **Greg Smithies** — 11/30/2018, 5:13 PM\n> So if we refuse to pay it, I don't think it's a massive issue for neuralink initially, though there is a lot on the line for openai's non-profit standing; so I'd expect them to get pretty nasty about it (ie probably willing to sue) if we didn't pay something that they could point their auditors to\n>\n> **Greg Smithies** — 11/30/2018, 5:18 PM\n> That said, almost 2/3rds of it is the rent number, which E is already paying the totality of in any case (I think?) so that wouldn't have to be an actual cash payment, but could be handled with some accounting memos and IOU's.\n>\n> **Jared Birchall** — 11/30/2018, 5:20 PM\n> Ok, thanks. I'll touch base with Chris to get his perspective.\n>\n> **Greg Smithies** — 11/30/2018, 5:20 PM\n> In other words - **we owe openai that rent, but then openai would technically owe that portion back to E**.\n>\n> **Greg Smithies** — 11/30/2018, 5:20 PM\n> Sounds good.\n\n## Commentary\n\nDX 849 is the cleanest **Pioneer Building accounting** receipt in the case. Defense uses it for the same proposition as [[DX-539|DX 539]]: the building was a Musk-personally-financed multi-tenant arrangement (Neuralink, OpenAI, SpaceX/Tesla), not an OpenAI-owned charitable asset — and the November 2018 reimbursement push from \"Chris\" (Chris Clark, OpenAI's CFO) was *driven by OpenAI's nonprofit auditors needing arms-length accounting*, not by an underlying obligation. The \"**we owe openai that rent, but then openai would technically owe that portion back to E**\" line is defense's textual punchline — the accounting is circular because Musk was paying the entire rent personally. **Plaintiffs' read** of the same text is, however, equally pointed: by November 2018, OpenAI's *own auditors* had concluded that without rent reimbursement OpenAI could not pass its nonprofit audits — i.e., the no-reimbursement Pioneer arrangement was a charitable-trust problem on the OpenAI side that OpenAI's professional auditors flagged. Cross-reference [[Key Themes]] § \"Pioneer Building,\" which collects the full sequence: [[DX-539|DX 539]] (June 2016 \"I don't want Sam on the lease\"), DX 600 (Jan 2017 \"Elon holds 100 percent of the power and authority\"), DX 849, DX 857 (Jan 2019, Musk countering OpenAI's $3.7M ask with $1.25M), and [[PX 103]] (July 2020 Clark email — plaintiffs' alleged 2020 § 17510.6 \"solicitation\"). Birchall on cross: \"**I'm not familiar with all of the legalities of donor advised funds**\" `(043026TT.txt:5007)`.\n\n---\n*See also:* [[Day 4|Day 4 digest]] · [[Pioneer Building]] · [[Jared Birchall]] · [[DX-539]] · [[PX 103]] · [[Statute of Limitations]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-853", "exhibit": "DX 853", "party": "Defense", "type": "Email thread", "admitted_trial_day": "Day 3 (April 29, 2026)", "uploaded_box_pt": "2026-04-29T15:36:47", "uploader": "Morrison Foerster", "pages": 2, "size_bytes": 400828, "source_pdf": "DX-853.pdf", "pdf_url": "https://media.mts-in.com/DX-853.pdf", "body_markdown": "# DX 853 — Dec 26–31, 2018 Musk \"zero percent / humanity's future is in the hands of Demis\" email\n\n> Musk's December 26, 2018 email to Sam Altman (cc Brockman, Sutskever, Zilis): \"**My probability assessment of OpenAI being relevant to DeepMind/Google without a dramatic change in execution and resources is 0%. Not 1%.**\" — used by defense as the cleanest evidence of the 2018 break and Musk's parallel-AGI pivot to Tesla, central to the [[Key Themes#Tesla / xAI as Musk's own AGI play (the parallel)|\"parallel AGI play\"]] frame.\n\n## Document type\n**Email thread, 2 pages, four messages.** From Elon Musk <erm@spacex.com> to Sam Altman <shga@ycombinator.com>, cc EMDesk <emdesk@spacex.com>, Greg Brockman <gdb@openai.com>, Ilya Sutskever <ilyasu@openai.com>, \"shivon@tesla.com\" <shivon@tesla.com>, subject \"Re: I feel I should reiterate,\" dated Mon, 31 Dec 2018 07:04:15 +0000. Production-stamped Confidential; Bates 2024MUSK-0009072–0009073. Document ID DX-0853.\n\n## Logistics\n- **Trial admission:** Day 3 (April 29, 2026) — Musk direct concludes; cross begins. Per the wiki TSV: \"**DX 853 — Dec 31 2018 zero-percent Musk email.**\" Cited in [[Key Themes]]: \"Musk to Altman/Brockman/Sutskever/Zilis, Dec. 31, 2018 (DX 853): 'My probability assessment of OpenAI being relevant to DeepMind/Google without a dramatic change in execution and resources is **zero percent: not 1 percent**.'\" (`042926TT.txt:5371`).\n- **Box upload:** 2026-04-29 15:36:47 PT — Day 3 late-afternoon batch (clustered with DX 827, DX 679, DX 862, DX 927).\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~392 KB, 2 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `853 .pdf` (note trailing space).\n\n## Transcribed text\n\n> **From:** Elon Musk <erm@spacex.com>\n> **To:** Sam Altman <shga@ycombinator.com>\n> **Cc:** EMDesk <emdesk@spacex.com>, Greg Brockman <gdb@openai.com>, Ilya Sutskever <ilyasu@openai.com>, Shivon Zilis <shivon@tesla.com>\n> **Subject:** Re: I feel I should reiterate\n> **Date:** Mon, 31 Dec 2018 07:04:15 -0000\n>\n> Ok\n\n> **From:** Sam Altman <shga@ycombinator.com>\n> **To:** Elon Musk <erm@spacex.com>\n> **Cc:** EMDesk <emdesk@spacex.com>, Greg Brockman <gdb@openai.com>, Ilya Sutskever <ilyasu@openai.com>, Shivon Zilis <shivon@tesla.com>\n> **Subject:** Re: I feel I should reiterate\n> **Date:** Dec 30, 2018, at 11:02 PM\n>\n> Let's do Sat in Puerto Rico!\n\n> **From:** Elon Musk <erm@spacex.com>\n> **To:** Sam Altman <shga@ycombinator.com>\n> **Cc:** EMDesk <emdesk@spacex.com>, Greg Brockman <gdb@openai.com>, Ilya Sutskever <ilyasu@openai.com>, Shivon Zilis <shivon@tesla.com>\n> **Subject:** Re: I feel I should reiterate\n> **Date:** Mon, Dec 31, 2018 at 4:01 AM\n>\n> I could meet on Wednesday in the Bay Area.\n>\n> Will also be at the AI safety conference on Saturday (Jan 5).\n>\n> OpenAI is not a serious counterweight to DeepMind/Google and will only get further behind. It is surprising that this isn't obvious to you.\n>\n> In general, always overestimate competitors. You are doing the opposite.\n\n> **From:** Sam Altman <shga@ycombinator.com>\n> **To:** Elon Musk <erm@spacex.com>\n> **Cc:** Greg Brockman <gdb@openai.com>, Ilya Sutskever <ilyasu@openai.com>, Shivon Zilis <shivon@tesla.com>\n> **Subject:** Re: I feel I should reiterate\n> **Date:** Dec 30, 2018, at 2:04 PM\n>\n> Can we meet the first week of January to discuss what we can do to increase that percentage?\n>\n> Speaking for myself, I believe we have a good plan, a credible path to get sufficient capital, but are not executing nearly fast enough.\n>\n> None of us want to be Bezos here!\n\n> **From:** Elon Musk <erm@spacex.com>\n> **To:** Sam Altman <shga@ycombinator.com>\n> **Cc:** Greg Brockman <gdb@openai.com>, Ilya Sutskever <ilyasu@openai.com>, Shivon Zilis <shivon@tesla.com>\n> **Subject:** I feel I should reiterate\n> **Date:** Wed, 26 Dec 2018 at 20:07\n>\n> My probability assessment of OpenAI being relevant to DeepMind/Google without a dramatic change in execution and resources is 0%. Not 1%. I wish it were otherwise.\n>\n> Even raising several hundred million won't be enough. This needs billions per year immediately or forget it.\n>\n> Unfortunately, humanity's future is in the hands of Demis.\n>\n> https://nyti.ms/2Rjetd3?smid=nytcore-ios-share\n>\n> And they are doing a lot more than this.\n>\n> OpenAI reminds me of Bezos and Blue Origin. They are hopelessly behind SpaceX and getting worse, but the ego of Bezos has him insanely thinking that they are not!\n>\n> I really hope I'm wrong.\n>\n> Elon\n\n## Commentary\n\nDX 853 is one of the defense's cleanest exhibits and also one of plaintiffs' most awkward. **For defense:** the December 26, 2018 message is admissible proof of the [[Key Themes#Musk's \"three phases\"|three-phases narrative]]'s second-phase break — by year-end 2018, fourteen months after the [[PX 157]] \"final straw\" exchange and four months after the [[DX-827]] purple-box term sheet, Musk was already publicly writing off OpenAI as irrelevant (\"**hopelessly behind**\") and pivoting all hope to Tesla as the AGI counterweight (\"**attempt that through Tesla instead**\" — DX 844, two months earlier). The \"**zero percent: not 1 percent**\" line is part of the defense's statute-of-limitations architecture: Musk knew the structure had broken by 2018 and did nothing for six years. **For plaintiffs:** Altman's prompt reply (\"Can we meet the first week of January to discuss what we can do to increase that percentage?\") and Musk's terse \"Ok\" to a meeting in Puerto Rico shows the relationship had not yet fully ruptured; Musk himself returned to \"phase one\" framing on the stand and said his **\"phase three\"** (looting / stole-a-charity) realization came in **late 2022 / 2023**, not 2018 (`042926TT.txt:7148` and the recross \"When was phase two?\" sequence in [[Key Themes]]). Note the email closes with \"I really hope I'm wrong\" — the line plaintiffs use to soften the apparent finality. Pairs naturally with [[DX-728]] (Teller's Dec 2017 \"**very low probability of OpenAI succeeding at AGI and we need to build up Tesla**\") and [[DX-758]] (Zilis's Feb 2018 nine-scenarios memo), forming the defense's complete 2017-2018 \"Musk was already pivoting to Tesla\" timeline.\n\n---\n*See also:* [[Day 3|Day 3 digest]] · [[Sam Altman]] · [[DX-728]] · [[DX-758]] · [[PX 157]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-857", "exhibit": "DX 857", "party": "Defense", "type": "Email", "admitted_trial_day": "Day 4 (April 30, 2026)", "uploaded_box_pt": "2026-04-30T15:13:15", "uploader": "Morrison Foerster", "pages": 2, "size_bytes": 372705, "source_pdf": "DX-857.pdf", "pdf_url": "https://media.mts-in.com/DX-857.pdf", "body_markdown": "# DX-857 — Jan 22, 2019 Musk-to-Birchall: \"Let's offer $250k and $1M to OpenAI\"\n\n> Musk's one-line counter to Birchall on the Pioneer Building reimbursement negotiation — atop Birchall's earlier balance-sheet update reporting OpenAI's escalating walk-down from $1.2M/$2.5M to $360k/$1.37M for 2017–2018.\n\n## Document type\n**Email thread, plain text, two-message reply chain.** Musk reply (\"erm@spacex.com\") to Jared Birchall (jared@excession.com), Re: Update, Tue, 22 Jan 2019 01:01:59 -0800; on top of Birchall's 12:12 AM same-day \"bump\" of his earlier Saturday Jan 19, 2019 1:48 AM \"Update\" message attaching `Balance sheet EM 1.18.19.xlsx`. Bates 2024MUSK-0008904–0008905. Substantial portions of the email body are redacted (other Musk-Foundation balance-sheet items).\n\n## Logistics\n- **Trial admission:** Day 4 (April 30, 2026) — Musk cross + redirect; Birchall direct + cross. Per [[Key Themes]]: \"DX 857 — Jan 2019 Pioneer Building reimbursement counter.\"\n- **Box upload:** 2026-04-30 15:13:15 PT — Day-4 mid-afternoon batch (clustered with DX 827, DX 1284).\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~373 KB, 2 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `857.pdf`.\n\n## Transcribed text\n\n> **From:** Elon Musk <erm@spacex.com>\n> **To:** Jared Birchall <jared@excession.com>\n> **Subject:** Re: Update\n> **Date:** Tue, 22 Jan 2019 01:01:59 -0800\n> **Importance:** Normal\n>\n> [redacted block]\n>\n> Let's offer **$250k and $1M to OpenAI**\n>\n> ---\n>\n> On Jan 22, 2019, at 12:12 AM, Jared Birchall <jared@excession.com> wrote:\n>\n> Bumping this up in your box. I've **bolded** any pressing questions.\n>\n> ---------- Forwarded message ---------\n> **From:** Jared Birchall <jared@excession.com>\n> **To:** Elon Musk <erm@spacex.com>\n> **Subject:** Update\n> **Date:** Sat, Jan 19, 2019 at 1:48 AM\n>\n> Elon,\n>\n> Attached is the updated balance sheet. [redacted]\n>\n> - **Pioneer Building - OpenAI/Neuralink Shared Expense** — OpenAI initially asked for a reimbursement of **$1.2 [million] for 2017 and $2.5 [million] for 2018**. That was rejected and they came back with **$740k for 2017 and $1.9M for 2018**. You rejected this. They sharpened the pencil and came back with **$360 [thousand] for 2017 and $1.37M for 2018**. I can explain the methodology if helpful. **Are these latest numbers more in-line with your expectations?**\n>\n> [redacted]\n>\n> Have a nice weekend,\n>\n> Jared\n>\n> <Balance sheet EM 1.18.19.xlsx>\n\n## Commentary\n\nDX 857 is the contemporaneous documentation of Musk's negotiating posture on Pioneer Building rent — central to plaintiffs' alternative § 17510.6 charitable-solicitation theory. The reimbursement walk-down is striking: OpenAI's opening ask was $3.7M (2017+2018 combined); after two Musk rejections, OpenAI sharpened to ~$1.73M; Musk's final counter here is **$1.25M total ($250k + $1M)** — the \"approximately two-thirds cut\" that Birchall confirmed on direct (`043026TT.txt:6452`). That this was negotiated through Birchall (Musk's personal banker at Excession, not OpenAI's lessor of record) reinforces plaintiffs' [[Day 4|Day 4]] theme that Musk personally controlled Pioneer Building economics through Musk Industries LLC, not as OpenAI's landlord ([[Key Themes]] Pioneer-Building section). Pairs with [[DX-539]] (\"I don't want Sam on the lease\"), [[DX-600]] (\"100 percent of the power and authority\"), [[DX-849]] (auditor pressure to charge Neuralink rent), and [[PX 103]] (the July 2020 \"happy to cover 100 percent of the rent\" email that plaintiffs use as the SOL-defeating 2020 solicitation).\n\n---\n\n*See also:* [[Day 4|Day 4 digest]] · [[Jared Birchall]] · [[Key Themes]] · [[DX-539]] · [[DX-600]] · [[DX-849]] · [[PX 103]] · [[Pioneer Building]]\n"}
{"exhibit_id": "DX-862", "exhibit": "DX 862", "party": "Defense", "type": "Email", "admitted_trial_day": "Day 3 (April 29, 2026)", "uploaded_box_pt": "2026-04-29T15:37:12", "uploader": "Morrison Foerster", "pages": 4, "size_bytes": 317026, "source_pdf": "DX-862.pdf", "pdf_url": "https://media.mts-in.com/DX-862.pdf", "body_markdown": "# DX 862 — March 6-8, 2019 Altman → Musk capped-profit announcement draft (Zilis fwd)\n\n> Sam Altman's draft of the imminent OpenAI LP capped-profit announcement, sent to Musk for review and feedback on March 6, 2019, and re-forwarded by Shivon Zilis on March 8 — the post that would publicly disclose the LP structure and explicitly note \"**Elon Musk left the board of OpenAI Nonprofit in February 2018 and is not involved with OpenAI LP.**\"\n\n## Document type\n**Email thread, four pages, two messages.** Zilis's March 8, 2019 3:24 AM PST forward (\"Fwd: OpenAI\") on top of Altman's March 6, 2019 3:12 PM PST original to Musk (cc Sam Teller, Shivon Zilis), subject \"OpenAI.\" The draft post text follows. Bates 2024MUSK-0006183–6186. Document ID DX-0862.\n\n## Logistics\n- **Trial admission:** Day 3 (April 29, 2026). Used by defense to establish that Musk was given a pre-publication preview of the LP announcement and the explicit \"Elon Musk left the board\" disclaimer — undermining plaintiffs' \"fraudulent concealment\" tolling theory.\n- **Box upload:** 2026-04-29 15:37:12 PT — Day 3 late-afternoon batch (clustered with DX 679, DX 827, DX 927).\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~310 KB, 4 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `862.pdf`. Marked Highly Confidential-AEO.\n\n## Transcribed text\n\n> **From:** Shivon Zilis <shivon@neuralink.com>\n> **To:** Elon Musk <erm@spacex.com>\n> **Subject:** Fwd: OpenAI\n> **Date:** Fri, 8 Mar 2019 03:24:17 -0800\n>\n> Elon,\n>\n> Bumping this back up. Did you have feedback for Altman / would you like to discuss this with him over the phone?\n>\n> In addition to his TL;DR, note the callout at the end clarifying your involvement:\n> - **Elon Musk left the board of OpenAI Nonprofit in February 2018 and is not involved with OpenAI LP**\n>\n> Altman said they did this in hopes of saving you media pain but it can say anything you'd like.\n>\n> Shivon\n>\n> Begin forwarded message:\n\n> **From:** Sam Altman <shga@ycombinator.com>\n> **To:** Elon Musk <erm@spacex.com>\n> **Cc:** Sam Teller <steller@spacex.com>, Shivon Zilis <shivon@neuralink.com>\n> **Subject:** OpenAI\n> **Date:** March 6, 2019 at 3:12:20 PM PST\n>\n> Elon —\n>\n> Here is a draft post we are planning for Monday. Anything to add/edit?\n>\n> **TL;DR:**\n> - We've created the capped-profit company and raised the first round, led by Reid and Vinod.\n> - We did this in a way where all investors are clear that they should never expect a profit (see purple box below).\n> - We made Greg chairman and me CEO of the new entity.\n> - We have tested this structure with potential next-round investors and they seem to like it.\n>\n> Speaking of the last point, we are now discussing a multi-billion dollar investment which I would like to get your advice on when you have time. Happy to come see you some time you are in the bay area.\n>\n> Sam\n>\n> ---\n>\n> **We've created OpenAI LP, a new \"capped-profit\" company that allows us to rapidly increase our investments in compute and talent while including checks and balances to actualize our mission.**\n>\n> Our mission is to ensure that artificial general intelligence (AGI) benefits all of humanity, primarily by attempting to build safe AGI and share the benefits with the world.\n>\n> Due to the exponential growth of compute investments in the field, we've needed to scale much faster than we'd planned when starting OpenAI. We expect to need to raise many billions of dollars in upcoming years for large-scale cloud compute, attracting and retaining talented people, and building AI supercomputers.\n>\n> We haven't been able to raise that much money as a nonprofit, and though we considered becoming a for-profit, we were afraid that doing so would mean giving up our mission. Instead, we created a new company, OpenAI LP, as a hybrid for-profit and nonprofit — which we are calling a \"capped-profit\" company.\n>\n> The fundamental idea of OpenAI LP is that investors and employees can get a fixed return if we succeed at our mission, which allows us to raise investment capital and attract employees with startup-like equity. But any returns beyond that amount — and if we are successful, we expect to generate orders of magnitude more value than we'd owe to people who invest in or work at OpenAI LP — are owned by the original OpenAI Nonprofit entity.\n>\n> Going forward (in this post and elsewhere), \"OpenAI\" refers to OpenAI LP (which now employs most of our staff), and the original entity is referred to as \"OpenAI Nonprofit\".\n>\n> [image: OpenAI team and their families at our November 2018 offsite.]\n>\n> **The mission comes first**\n>\n> We've designed OpenAI LP to put our overall mission — ensuring the creation and adoption of safe and beneficial AGI — over generating returns for investors.\n>\n> To minimize conflicts of interest with the mission, OpenAI LP's primary fiduciary obligation is to advance the aims of the OpenAI Charter, and the company is controlled by OpenAI Nonprofit's board. All investors and employees sign agreements that OpenAI LP's obligation to the Charter always comes first, even at the expense of some or all of their financial stake.\n>\n> [image: Big purple box]\n> *Our employee and investor paperwork starts like this. The general partner refers to OpenAI Nonprofit (whose official name is \"OpenAI Inc\"); limited partners refers to investors and employees.*\n>\n> Only a minority of board members can hold financial stakes in the partnership. Furthermore, only board members without such stakes are allowed to vote on decisions where the interests of limited partners and the nonprofit's mission may conflict — including any decisions about making payouts to investors and employees.\n>\n> [image: Corporate structure]\n> *Another provision from our paperwork specifies that the nonprofit retains control. (The paperwork uses OpenAI LP's official name \"OpenAI, L.P.\".)*\n>\n> As mentioned above, economic returns for investors and employees are capped (with the cap negotiated in advance on a per-limited partner basis). Any excess returns are owned by the nonprofit. Our goal is to ensure that most of the value we create if successful is returned to the world, so we think this is an important first step. Returns for our first round of investors are capped to 100x their investment, and we expect this multiple to be lower for future rounds.\n>\n> **What OpenAI does**\n>\n> Our day-to-day work remains the same. Today, we believe we can build the most value by focusing exclusively on developing new AI technologies, not commercial products. Our structure gives us flexibility for how to make money in the long term, but we hope to figure that out only once we've created safe AGI (though we're open to non-distracting revenue sources such as licensing in the interim).\n>\n> OpenAI LP currently employs around 100 people organized into three main areas: capabilities (advancing what AI systems can do), safety (ensuring those systems are aligned with human values), and policy (ensuring appropriate governance for such systems). OpenAI Nonprofit governs OpenAI LP, runs educational programs such as Scholars and Fellows, and hosts policy initiatives. OpenAI LP is continuing (at increased pace and scale) the development roadmap started at OpenAI Nonprofit, which has yielded breakthroughs in reinforcement learning, robotics, and language.\n>\n> **Safety**\n>\n> We are concerned about AGI's potential to cause rapid change, whether through machines pursuing goals misspecified by their operator, malicious humans subverting deployed systems, or an out-of-control economy that grows without resulting in improvements to human lives. As described in our Charter, we are willing to merge with a value-aligned organization (even if it means reduced or zero payouts to investors) to avoid a competitive race which would make it hard to prioritize safety.\n>\n> **Who's involved**\n> - OpenAI Nonprofit's board consists of OpenAI LP employees Greg Brockman (Chairman & CTO), Ilya Sutskever (Chief Scientist), and Sam Altman (CEO), and non-employees Adam D'Angelo, Holden Karnofsky, Reid Hoffman, Sue Yoon, and Tasha McCauley.\n> - **Elon Musk left the board of OpenAI Nonprofit in February 2018 and is not involved with OpenAI LP.**\n> - Our investors include Reid Hoffman and Khosla Ventures.\n>\n> We are traveling a hard and uncertain path, but we have designed our structure to help us positively affect the world should we succeed in creating AGI. If you'd like to help us make this mission a reality, we're hiring :)!\n\n## Commentary\n\nDX 862 is the linchpin of defense's \"no fraudulent concealment\" argument. Altman sent the entire draft to Musk for advance review on March 6, 2019, two days before the public announcement. The post itself: (i) discloses the **capped-profit structure** with explicit naming of the **\"purple box\"** ([[DX 827]]); (ii) names the first-round investors (Reid Hoffman and Vinod Khosla); (iii) reveals the 100x cap (\"**Returns for our first round of investors are capped to 100x their investment**\"); (iv) tells Musk that Altman wanted his \"advice\" on a then-pending \"**multi-billion dollar investment**\" — the soon-to-be-announced Microsoft $1B; and (v) carries the explicit Musk-status disclaimer \"**Elon Musk left the board of OpenAI Nonprofit in February 2018 and is not involved with OpenAI LP**\" — which Altman framed (per Zilis) as drafted \"**in hopes of saving you media pain but it can say anything you'd like**.\" Defense uses this to argue (a) Musk had actual notice of the capped-profit structure, the 100x cap, the new investors, and the impending Microsoft round — well outside any later limitations period; and (b) Altman invited Musk's edits, the opposite of concealment. The \"**OpenAI LP's primary fiduciary obligation is to advance the aims of the OpenAI Charter**\" line ties this exhibit into [[PX 24]]. Plaintiffs read the same post as the moment OpenAI's commercial pivot was sanitized for the public — and the \"we are willing to merge with a value-aligned organization\" line as the rhetorical hedge that has not aged well.\n\n---\n*See also:* [[Day 3|Day 3 digest]] · [[DX 827]] · [[PX 24]] · [[Sam Altman]] · [[Shivon Zilis]] · [[Statute of Limitations]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-863", "exhibit": "DX 863", "party": "Defense", "type": "Text messages", "admitted_trial_day": "Day 3 (April 29, 2026)", "uploaded_box_pt": "2026-04-29T15:37:12", "uploader": "Morrison Foerster", "pages": 1, "size_bytes": 150154, "source_pdf": "DX-863.pdf", "pdf_url": "https://media.mts-in.com/DX-863.pdf", "body_markdown": "# DX 863 — March 10, 2019 Altman → Musk \"announce the new openai structure tomorrow\" texts\n\n> Sam Altman's March 10, 2019 4:33 AM text giving Musk a heads-up the day before OpenAI publicly announced the LP structure — explicitly offering to \"**describe your past involvement in whatever way you want**\" and inviting feedback, plus \"**some mild demis updates to share**.\"\n\n## Document type\n**Text messages, Cellebrite-style \"SMS-MMS-Chats\" extract.** One page, four messages, two participants (Sam Altman and Elon Musk, handles redacted). Conversation name `70ea4d289b4478b885c6e84501f19aa301da11cc2c4d8c439eb52dda0db3a333`. Timezone UTC-0700. Single date: 2019/03/10. Bates 2024MUSK-0014335. Document ID DX-0863.\n\n## Logistics\n- **Trial admission:** Day 3 (April 29, 2026). No prior wiki reference; admission day inferred from Box upload date.\n- **Box upload:** 2026-04-29 15:37:12 PT — Day 3 late-afternoon defense batch (clustered with [[DX 686]], [[DX 749]], [[DX 773]]).\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~147 KB, 1 page.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `863 .pdf` (note trailing space).\n\n## Transcribed text\n\n> **Platform:** SMS-MMS-Chats\n> **Name:** 70ea4d289b4478b885c6e84501f19aa301da11cc2c4d8c439eb52dda0db3a333\n> **Timezone:** UTC-0700\n> **Participants:** Sam Altman [redacted]; Elon Musk [redacted]\n>\n> **Date: 2019/03/10**\n>\n> **Sam Altman** — 04:33:00 am\n> hey elon, do you have a few mins to talk? we are looking to announce the new openai structure tomorrow and i wanted to make sure we describe your past involvement in whatever way you want, and would also welcome any other feedback you have! also have some mild demis updates to share.\n>\n> **Elon Musk** — 10:15:00 am\n> Sure. I have several calls to do this evening, but free on the late side.\n>\n> **Sam Altman** — 10:16:00 am\n> works for me\n>\n> **Sam Altman** — 03:53:00 pm\n> still good to talk tonight?\n>\n> **Elon Musk** — 03:53:00 pm\n> Yeah, good to talk at 11\n\n## Commentary\n\nDX 863 is a small but pointed defense exhibit on the timing question. The OpenAI LP structure was publicly announced on **March 11, 2019**. The day before, Altman texted Musk to give him advance notice and to offer Musk editorial control over how the press release described his \"past involvement.\" Defense uses this for the [[Key Themes#The 2018 \"purple box\" term sheet|disclosure / acquiescence]] argument: Musk was not surprised by the LP announcement six months later — he was personally consulted the day before, did not object, and did not sue. The \"**describe your past involvement in whatever way you want**\" line is the cleanest contemporaneous evidence that Altman's relationship to Musk's role was cooperative rather than concealing. The \"**mild demis updates**\" line refers to Demis Hassabis / DeepMind — defense reads it as evidence that the Musk-Altman channel on AI competitive intelligence was still open in March 2019. Cross-reference: [[PX 236]] / [[DX 828]] (the August 2018 LP term sheet — six months prior); [[PX 105]] (the October 2020 \"appear hypocritical\" texts — eighteen months later, when the relationship had cooled).\n\n---\n*See also:* [[Day 3|Day 3 digest]] · [[Sam Altman]] · [[PX 236]] · [[DX 828]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-869", "exhibit": "DX 869", "party": "Defendants", "type": "Text messages", "admitted_trial_day": "Day 3 (April 29, 2026)", "uploaded_box_pt": "2026-04-29T15:38:00", "uploader": "Morrison Foerster", "pages": 1, "size_bytes": 74752, "source_pdf": "DX-869.pdf", "pdf_url": "https://media.mts-in.com/DX-869.pdf", "body_markdown": "# DX 869 — April 3, 2019 Sam Altman text to Elon Musk: \"do you have a few mins to talk about the microsoft/openai investment?\"\n\n> A single, one-line SMS sent by Sam Altman to Elon Musk at 8:22 AM PDT on April 3, 2019, about three months after Musk's public departure from the OpenAI board (Feb 2018) and roughly four months before Microsoft's July 22, 2019 announcement of its $1B investment in OpenAI, L.P. The message is the affirmative paper trail that Altman attempted to consult Musk on the Microsoft transaction in real time.\n\n## Document type\n**Text messages.** A single SMS in a two-participant chat (SMS-MMS-Chats platform). Participants: Sam Altman and Elon Musk. Bates 2024MUSK-0014343.\n\n## Logistics\n- **Trial admission:** Day 3 (April 29, 2026) — Musk direct concludes; cross begins. Used by defense on cross to undermine Musk's claim that he was \"shut out\" of the for-profit conversion and the Microsoft deal — Altman's text shows Altman reaching out to Musk *directly* on the Microsoft-OpenAI investment two months after Musk left the board. Pairs with [[DX 691]] (Sept 2017 \"Ok by me\" on creating the B corp), [[DX 806]] (Apr 2018 \"Ok by me\" on capped-profit), and the Microsoft thread that becomes [[PX 203]] § Microsoft Convertible LP Interest in 2021.\n- **Box upload:** 2026-04-29 15:38:00 PT — Day 3 mid-afternoon batch.\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~73 KB, 1 page.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `869 .pdf`.\n\n## Transcribed text\n\n> **Platform:** SMS-MMS-Chats\n> **Name:** 60e2980d976c4e8e4a668494f77bb2865445d6f2911beaaf97c02720fac8df99\n> **Timezone:** UTC-0700\n>\n> Participants: Sam Altman, Elon Musk\n>\n> ---\n>\n> **2019/04/03**\n>\n> **Sam Altman** — 08:22:00 am\n>\n> do you have a few mins to talk about the microsoft/openai investment?\n\n[The exhibit reproduces only Altman's outgoing message; no reply from Musk appears in the produced extract.]\n\n## Commentary\n\nDX 869 is a one-sentence exhibit doing one job. **Defense's read:** in early April 2019 — about a week after the **OpenAI, L.P.** capped-profit structure was unveiled publicly (March 11, 2019, see [[Key Themes]] § \"The capped-profit conversion\") and roughly fifteen weeks before the Microsoft $1B announcement (July 22, 2019) — Altman texted Musk directly to consult him on the **Microsoft/OpenAI investment**. That cuts against plaintiffs' \"stole a charity\" framing in two ways: (i) it shows Altman treating Musk as an interlocutor on the very transaction plaintiffs say was concealed from him, and (ii) the absence of a written reply (combined with Musk's general disengagement after his Feb 2018 board departure) is consistent with the inference that Musk's silence — not Altman's secrecy — drove his exclusion from later for-profit transactions. **Plaintiffs' read:** \"do you have a few mins to talk\" is a courtesy gesture about a deal *already in the works*, not a meaningful consultation; the actual substance of the Microsoft Convertible LP Interest economics — the **$20B cap on Microsoft's 75% share, then nonprofit 100B**, the **AGI carveout**, and the **2032 IP termination** (see [[PX 203]] § 5.1(b)(ii) and § 6.4) — was negotiated over the following months without Musk and on terms he would later challenge as having \"captured\" the nonprofit. The exhibit is also a contemporaneous fixed point on the timeline: by April 3, 2019, the Microsoft deal was sufficiently advanced that Altman wanted to \"talk\" about it, three and a half months before the deal was announced.\n\n---\n*See also:* [[Day 3|Day 3 digest]] · [[PX 203]] · [[DX 691]] · [[DX 806]] · [[Sam Altman]] · [[Elon Musk]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-877", "exhibit": "DX 877", "party": "Defense (OpenAI/MS)", "type": "Blog post", "admitted_trial_day": "Day 8 (May 6, 2026)", "uploaded_box_pt": "2026-05-06T14:57:01", "uploader": "Morrison Foerster", "pages": 3, "size_bytes": 2716063, "source_pdf": "DX-877.pdf", "pdf_url": "https://media.mts-in.com/DX-877.pdf", "body_markdown": "# DX 877 — July 22, 2019 OpenAI blog post: \"Microsoft invests in and partners with OpenAI to support us building beneficial AGI\"\n\n> OpenAI's public-facing **July 22, 2019** company blog post announcing the **$1 billion Microsoft investment**, the Azure exclusive-cloud-provider relationship, and the framing that Microsoft would become OpenAI's \"preferred partner for commercializing\" pre-AGI technologies — defense's [[Key Themes#\"Captured by Microsoft\" (statute-of-limitations fight)|statute-of-limitations notice]] artifact.\n\n## Document type\n**Blog post / press release, public OpenAI company-page article**, dated July 22, 2019, illustrated with a Justin Jay Wang × DALL·E header image and a July 2019 OpenAI team-offsite group photo. Bates OPENAI_MUSK00037558–37560.\n\n## Logistics\n- **Trial admission:** Day 8 (May 6, 2026) — Murati live; Zilis live; Toner video begins. Used by defense as part of the public-notice cluster for the [[Key Themes#\"Captured by Microsoft\" (statute-of-limitations fight)|\"captured by Microsoft\" / SOL]] sequence and the [[Key Themes#The 2018 \"purple box\" term sheet|\"purple box\" / capped-profit notice]] timeline — published roughly eleven months after [[DX-827|the August 2018 LP term sheet]] and fourteen months before Musk's Sept 24, 2020 \"captured by Microsoft\" tweet (PX 251).\n- **Box upload:** 2026-05-06 14:57:01 PT — Day 8 mid-afternoon batch (clustered with [[DX-716]], [[DX-754]], [[DX-758]], [[DX-835]]).\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~2.59 MB, 3 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `0877.pdf`.\n\n## Transcribed text\n\n> **July 22, 2019 — Company**\n>\n> # Microsoft invests in and partners with OpenAI to support us building beneficial AGI\n>\n> *Illustration: Justin Jay Wang × DALL·E*\n>\n> Microsoft is investing $1 billion in OpenAI to support us building artificial general intelligence (AGI) with widely distributed economic benefits. We're partnering to develop a hardware and software platform within Microsoft Azure which will scale to AGI. We'll jointly develop new Azure AI supercomputing technologies, and Microsoft will become our exclusive cloud provider—so we'll be working hard together to further extend Microsoft Azure's capabilities in large-scale AI systems.\n>\n> Each year since 2012, the world has seen a new step function advance in AI capabilities. Though these advances are across very different fields like vision (2012), simple video games (2013), machine translation (2014), complex board games (2015), speech synthesis (2016), image generation (2017), robotic control (2018), and writing text (2019), they are all powered by the same approach: innovative applications of deep neural networks coupled with increasing computational power. But still, AI system building today involves a lot of manual engineering for each well-defined task.\n>\n> In contrast, an AGI will be a system capable of mastering a field of study to the world-expert level, and mastering more fields than any one human—like a tool which combines the skills of Curie, Turing, and Bach. An AGI working on a problem would be able to see connections across disciplines that no human could. We want AGI to work with people to solve currently intractable multi-disciplinary problems, including global challenges such as climate change, affordable and high-quality healthcare, and personalized education. We think its impact should be to give everyone economic freedom to pursue what they find most fulfilling, creating new opportunities for all of our lives that are unimaginable today.\n>\n> *[Photo caption: OpenAI team and their families at our July 2019 offsite.]*\n>\n> OpenAI is producing a sequence of increasingly powerful AI technologies, which requires a lot of capital for computational power. The most obvious way to cover costs is to build a product, but that would mean changing our focus. Instead, we intend to license some of our pre-AGI technologies, with Microsoft becoming our preferred partner for commercializing them.\n>\n> We believe that the creation of beneficial AGI will be the most important technological development in human history, with the potential to shape the trajectory of humanity. We have a hard technical path in front of us, requiring a unified software engineering and AI research effort of massive computational scale, but technical success alone is not enough. To accomplish our mission of ensuring that AGI (whether built by us or not) benefits all of humanity, we'll need to ensure that AGI is deployed safely and securely; that society is well-prepared for its implications; and that its economic upside is widely shared. If we achieve this mission, we will have actualized Microsoft and OpenAI's shared value of empowering everyone.\n\n## Commentary\n\nDX 877 is the public-record bookend to the 2019 Microsoft round Robert Wu testified about on Day 7 — **$1 billion in, $20 billion target redemption** ([[Key Themes#\"$250 billion\" — the target redemption stack]]) — and is defense's clean SOL artifact: Musk had public, unambiguous notice in **July 2019** that OpenAI had taken $1B from Microsoft, made Azure its exclusive cloud, and would \"license some of our pre-AGI technologies, with Microsoft becoming our preferred partner for commercializing them.\" That is the structural fact the [[PX-157|Sept 2017 final-straw email]] and the founders' \"Honest Thoughts\" letter had warned about, and it sits over fourteen months before Musk's Sept 24, 2020 \"**captured by Microsoft**\" tweet that defense uses to start the SOL clock. Plaintiffs' read is the inverse: the post studiously preserves the 2018 Charter's mission language (\"AGI… benefits all of humanity,\" \"widely distributed economic benefits\") and explicitly carves AGI out of the commercialization arrangement (\"license some of our **pre-AGI** technologies\"), exactly the AGI-carveout that the Day 7 [[Key Themes#The Watershed MOU — and the AGI carveout dissolving|Watershed MOU]] now threatens to dissolve. Plaintiffs will say the breach plaintiffs sue over is the *cumulative* dismantling of that carveout, not its 2019 announcement. See [[Day 8|Day 8 digest]] and [[Key Themes]].\n\n---\n*See also:* [[Day 8|Day 8 digest]] · [[Key Themes]] · [[Microsoft]] · [[OpenAI Charter]] · [[DX-827]]\n"}
{"exhibit_id": "DX-899", "exhibit": "DX 899", "party": "Defense", "type": "Blog post", "admitted_trial_day": "Day 7 (May 5, 2026)", "uploaded_box_pt": "2026-05-05T14:27:02", "uploader": "Morrison Foerster", "pages": 8, "size_bytes": 2932184, "source_pdf": "DX-899.pdf", "pdf_url": "https://media.mts-in.com/DX-899.pdf", "body_markdown": "# DX 899 — June 11, 2020 OpenAI blog post: \"OpenAI API\" launch (with FAQ \"Why did OpenAI choose to release an API instead of open-sourcing the models?\")\n\n> The June 11, 2020 OpenAI blog post launching the **OpenAI API** (the GPT-3 commercial product), with a self-explanatory FAQ section: \"**Why did OpenAI choose to release an API instead of open-sourcing the models?**\" — three reasons: commercialization funds research/safety, the models are too large for non-corporates to run, and an API allows OpenAI to \"**respond to misuse**\" in ways open-source release would not.\n\n## Document type\n**Blog post** — OpenAI.com Product post dated June 11, 2020, updated September 18, 2020. Multi-page PDF screenshot capture with API code-sample and demo screenshots. Bates OPENAI_MUSK00039684–00039691.\n\n## Logistics\n- **Trial admission:** Day 7 (May 5, 2026) — Brockman defense direct + Mr. Molo recross + DX 1252; Wu deposition. Used by defense in the [[Key Themes]] §\"The 'OPEN' in OpenAI\" line of attack on plaintiffs' \"open\" = \"open source\" theory.\n- **Box upload:** 2026-05-05 14:27:02 PT — Day 7 mid-afternoon Defense batch.\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~2.9 MB, 8 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `0899.pdf`.\n\n## Transcribed text\n\n> **June 11, 2020 — Product**\n>\n> # OpenAI API\n>\n> We're releasing an API for accessing new AI models developed by OpenAI.\n>\n> [Image of orange/blue grid pattern]\n>\n> We're releasing an API for accessing new AI models developed by OpenAI. Unlike most AI systems which are designed for one use-case, the API today provides a general-purpose \"text in, text out\" interface, allowing users to try it on virtually any English language task.\n>\n> You can now request access in order to integrate the API into your product, develop an entirely new application, or help us explore the strengths and limits of this technology.\n>\n> [bash terminal screenshot showing curl example: model \"davinci\", prompt \"Q: What is human life expectancy in the United States? … Q: What is the meaning of life?\", response: \"A: The meaning of life is 42.\"]\n>\n> Given any text prompt, the API will return a text completion, attempting to match the pattern you gave it. You can \"program\" it by showing it just a few examples of what you'd like it to do; its success generally varies depending on how complex the task is. The API also allows you to hone performance on specific tasks by training on a dataset (small or large) of examples you provide, or by learning from human feedback provided by users or labelers.\n>\n> We've designed the API to be both simple for anyone to use but also flexible enough to make machine learning teams more productive. … **Today the API runs models with weights from the GPT-3 family with many speed and throughput improvements.**\n>\n> [Excel + OpenAI API \"Tabulate\" screenshot, Wikipedia screenshot, terminal-based natural-language-shell screenshot]\n>\n> The field's pace of progress means that there are frequently surprising new applications of AI, both positive and negative. We will terminate API access for obviously harmful use-cases, such as harassment, spam, radicalization, or astroturfing. … We are launching today in a private beta rather than general availability, building tools to help users better control the content our API returns, and researching safety-relevant aspects of language technology …\n>\n> In addition to being a revenue source to help us cover costs in pursuit of our mission, the API has pushed us to sharpen our focus on general-purpose AI technology — advancing the technology, making it usable, and considering its impacts in the real world. … Interested in exploring the API? Join companies like Algolia, Quizlet, and Reddit, and researchers at institutions like the Middlebury Institute in our private beta.\n>\n> ## Frequently asked questions\n>\n> **Why did OpenAI decide to release a commercial product?**\n> Ultimately, what we care about most is ensuring artificial general intelligence benefits everyone. We see developing commercial products as one of the ways to make sure we have enough funding to succeed.\n> We also believe that safely deploying powerful AI systems in the world will be hard to get right. In releasing the API, we are working closely with our partners to see what challenges arise when AI systems are used in the real world. This will help guide our efforts to understand how deploying future AI systems will go, and what we need to do to make sure they are safe and beneficial for everyone.\n>\n> **Why did OpenAI choose to release an API instead of open-sourcing the models?**\n> There are three main reasons we did this.\n> First, **commercializing the technology helps us pay for our ongoing AI research, safety, and policy efforts**.\n> Second, many of the models underlying the API are very large, taking a lot of expertise to develop and deploy and making them very expensive to run. This makes it hard for anyone except larger companies to benefit from the underlying technology. We're hopeful that the API will make powerful AI systems more accessible to smaller businesses and organizations.\n> Third, the API model allows us to more easily respond to misuse of the technology. Since it is hard to predict the downstream use cases of our models, **it feels inherently safer to release them via an API and broaden access over time, rather than release an open source model where access cannot be adjusted if it turns out to have harmful applications.**\n>\n> **What specifically will OpenAI do about misuse of the API, given what you've previously said about GPT-2?**\n> With GPT-2, one of our key concerns was malicious use of the model (e.g., for disinformation), which is difficult to prevent once a model is open sourced. For the API, we're able to better prevent misuse by limiting access to approved customers and use cases. We have a mandatory production review process before proposed applications can go live. … We terminate API access for use cases that are found to cause (or are intended to cause) physical, emotional, or psychological harm to people, including but not limited to harassment, intentional deception, radicalization, astroturfing, or spam …\n>\n> **How will OpenAI mitigate harmful bias and other negative effects of models served by the API?**\n> Mitigating negative effects such as harmful bias is a hard, industry-wide issue that is extremely important. As we discuss in the GPT-3 paper and model card, our API models do exhibit biases that will be reflected in generated text. …\n>\n> *Updated September 18, 2020*\n\n## Commentary\n\nDX 899 is defense's public-facing rebuttal to the [[Key Themes]] §\"The 'OPEN' in OpenAI\" theme. The blog post is the **OpenAI-stated explanation** for why a commercial API replaced open-source release, written in OpenAI's own voice three months before the [[PX 921]] September 2020 GPT-3 / Microsoft licensing announcement that triggered Musk's \"captured by Microsoft\" tweet. Defense's argument is two-fold: (i) the public, contemporaneous justification — *commercialization funds research and safety; large models cannot be run by anyone but large companies; API release makes misuse-control feasible* — is a coherent mission-aligned rationale, not concealment; and (ii) Musk took no action against this announcement at the time, which feeds the SOL theory (see [[Key Themes]] §\"Captured by Microsoft (statute-of-limitations fight)\"). Plaintiffs' counter is the judge's Day 2 admonition to OpenAI's counsel about not taking inconsistent positions in front of the PTO on whether \"open\" meant \"open source\" (see [[Key Themes]] §\"The 'OPEN' in OpenAI\"), and Musk's Day 2 testimony that \"**the 'OPEN' in OpenAI represents open source.**\" Pair with [[PX 921]] for the September 2020 GPT-3 / Microsoft announcement and with [[PX 244]] for the May 2019 Murati ↔ Microsoft Odyssey-JDCA negotiation behind both.\n\n---\n*See also:* [[Day 7|Day 7 digest]] · [[PX 244]] · [[PX 921]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-900", "exhibit": "DX 900", "party": "Defense", "type": "Text messages", "admitted_trial_day": "Day 8 (May 6, 2026)", "uploaded_box_pt": "2026-05-06T14:57:00", "uploader": "Morrison Foerster", "pages": 2, "size_bytes": 95753, "source_pdf": "DX-900.pdf", "pdf_url": "https://media.mts-in.com/DX-900.pdf", "body_markdown": "# DX 900 — July 22, 2019 Zilis → Sam Altman, \"Congrats! … impressive deal by you guys\"\n\n> A two-page text-thread extract from July 22, 2019 — Shivon Zilis to Sam Altman shortly after OpenAI's announced **$1B Microsoft investment**: \"**Congrats! I'm sure you're in touch with him too, but when I'd pinged E the blog post he mentioned he thought it was an impressive deal by you guys**\" — followed by \"Also, you have unreasonably good posture. I need to learn.\"\n\n## Document type\n**Text messages, screenshot transcript.** Two pages (cover sheet + one page of conversation). One chat, 4 messages, 4 participants. The same message text appears twice (4:07–4:08 PM and 11:43–11:44 PM) consistent with a forwarded-or-replicated thread. Sender labeled \"shivon zilis\" with handle redacted; recipients include Sam Altman (two handles redacted). Bates OPENAI_MUSK00018000–18001.\n\n## Logistics\n- **Trial admission:** Day 8 (May 6, 2026 — Murati live; Zilis live; Toner video begins). No prior wiki citation; admission day inferred from Box upload and Zilis's live testimony.\n- **Box upload:** 2026-05-06 14:57:00 PT — Day 8 mid-afternoon defense batch (clustered with DX 619, DX 664, DX 724, DX 757, DX 824).\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~93 KB, 2 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `0900.pdf`.\n\n## Transcribed text\n\n> **Date Range:** 7/22/2019 — **Total Messages:** 4 — **Participants:** 4 (Sam Altman [two handles], shivon zilis)\n> [Times in GMT +00:00]\n>\n> **shivon zilis** — 7/22/2019, 4:07 PM\n> Congrats! I'm sure you're in touch with him too, but when I'd pinged E the blog post he mentioned he thought it was an impressive deal by you guys\n>\n> **shivon zilis** — 7/22/2019, 4:08 PM\n> Also, you have unreasonably good posture. I need to learn\n>\n> **shivon zilis** — 7/22/2019, 11:43 PM\n> Congrats! I'm sure you're in touch with him too, but when I'd pinged E the blog post he mentioned he thought it was an impressive deal by you guys\n>\n> **shivon zilis** — 7/22/2019, 11:44 PM\n> Also, you have unreasonably good posture. I need to learn\n\n## Commentary\n\nDX 900 is a tiny but tactical exhibit. Defense uses it to nail down **Musk's contemporaneous reaction to the announcement of the OpenAI–Microsoft $1B partnership** (July 22, 2019, the deal that became the **2019 Microsoft round** with **$1B in / $20B target redemption** — see [[Key Themes]] § \"Day 7 — Microsoft economics\"). Through Zilis as messenger, Musk's response to the very deal his own [[PX 251|September 24, 2020 tweet]] would later call \"OpenAI is essentially captured by Microsoft\" was, in real time: \"**impressive deal by you guys**.\" The exhibit is part of defense's **statute-of-limitations** package: if Musk understood the Microsoft architecture in July 2019 well enough to call it \"impressive,\" the SOL on his current \"captured by Microsoft\" theory of breach starts running then, not in 2023. The \"**I need to learn**\" / \"**unreasonably good posture**\" line is incidental tone, included in the produced transcript as authenticity. The duplicate timestamps (4:07/4:08 PM and 11:43/11:44 PM) appear to reflect a forwarded thread or a Cellebrite duplicate-extraction artifact rather than two separate sends.\n\n---\n*See also:* [[Day 8|Day 8 digest]] · [[Shivon Zilis]] · [[Sam Altman]] · [[Microsoft]] · [[PX 251]] · [[Statute of Limitations]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-925", "exhibit": "DX 925", "party": "Defense", "type": "Text messages", "admitted_trial_day": "Day 8 (May 6, 2026)", "uploaded_box_pt": "2026-05-06T14:57:00", "uploader": "Morrison Foerster", "pages": 2, "size_bytes": 82945, "source_pdf": "DX-925.pdf", "pdf_url": "https://media.mts-in.com/DX-925.pdf", "body_markdown": "# DX 925 — Oct 18, 2020 Altman → Zilis text: \"still think a good idea for me to ping elon for advice on the MSFT fundraise?\"\n\n> Two-message Cellebrite-style \"Short Message Report\" capturing Sam Altman texting Shivon Zilis on October 18, 2020 — three months after the GPT-3 paper and one month after Musk's Sept 24, 2020 \"**captured by Microsoft**\" tweet — to ask whether he should \"**ping elon for advice on the MSFT fundraise**.\"\n\n## Document type\n**Text-message extraction report (Cellebrite-format), 2 pages.** \"Short Message Report\" cover sheet (Conversations: 1; Total Messages: 2; Participants: 4; Date Range: 10/18/2020) followed by chronological message page. Phone numbers redacted. From: Sam Altman; recipient on this conversation: shivon zilis. Production-stamped CONFIDENTIAL; Bates OPENAI_MUSK00018211–18212. Document ID DX-0925.\n\n## Logistics\n- **Trial admission:** Day 8 (May 6, 2026) — Murati live; Zilis live; Toner video begins. Used during Zilis's live testimony on the Musk/Microsoft 2020 thread alongside Musk's Sept 24, 2020 \"**captured by Microsoft**\" tweet (PX 251) — see [[Key Themes#\"Captured by Microsoft\" (statute-of-limitations fight)]] and the Zilis quote \"**That being terrifying because it was just not the thing that we'd been fighting so hard for**\" (`5/6/2026 Testimony @ ~11:14 PT`).\n- **Box upload:** 2026-05-06 14:57:00 PT — Day 8 mid-afternoon batch (clustered with DX 643, DX 707, DX 728, DX 758, DX 827).\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~81 KB, 2 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `0925.pdf`.\n\n## Transcribed text\n\n> **Short Message Report**\n> Conversations: 1 | Participants: 4\n> Total Messages: 2 | Date Range: 10/18/2020\n>\n> **Outline of Conversations**\n>\n> CHAT — 96518_003.06.0000001 — 02011 — 2020/10/18 · 2 messages on 10/18/2020\n> Participants: Sam Altman <[redacted]> · Sam Altman <[redacted]> · Sam Altman <[redacted]> · shivon zilis <[redacted]>\n>\n> ---\n>\n> **Messages in chronological order** (times shown in GMT +00:00)\n>\n> CHAT — 96518_003.06.0000001 — 02011 — 2020/10/18\n>\n> **SA — Sam Altman** *(10/18/2020, 4:27 PM)*\n> still think a good idea for me to ping elon for advice on the MSFT fundraise? if so this is about the time for me to do it\n>\n> **SA — Sam Altman** *(10/18/2020, 4:27 PM)*\n> (i am positively inclined)\n\n## Commentary\n\nDX 925 is a tiny exhibit doing an outsized job in defense's [[Key Themes#\"Captured by Microsoft\" (statute-of-limitations fight)|statute-of-limitations]] story. October 18, 2020 places it **24 days after Musk's \"OpenAI is essentially captured by Microsoft\" tweet** and roughly nine months after Microsoft's first $1B 2019 round had closed — Altman is asking Zilis (then Musk's information conduit and an OpenAI nonprofit-board-bound figure) whether to *seek Musk's advice* on the next Microsoft fundraise. For defense, that's a contemporaneous fact undercutting the plaintiffs' \"concealment\" theory: the Altman side wasn't hiding the Microsoft fundraise from Musk; Altman was actively considering looping him in. It also pairs with Musk's own cross-examination concession that \"**Sam Altman immediately reached out to reassure me that OpenAI was on its — was staying on mission**\" after the September 2020 tweet (`042926TT.txt:2194`) — reinforcing the defense story that Musk was *informed* in 2020 and *did nothing* until 2024. For plaintiffs, the message reads in a different direction: Altman is asking Zilis whether *now* — while Musk is publicly tweeting about the Microsoft capture — is a strategic moment to engage him, suggesting the relationship is being managed rather than transparent. The \"(i am positively inclined)\" parenthetical is the kind of phrasing the plaintiffs will weave into the [[Key Themes#\"Pageant of hypocrisy\" / Tale of two Elons|Brockman-journal \"two truths\"]] thread that consumed Day 7.\n\n---\n*See also:* [[Day 8|Day 8 digest]] · [[Sam Altman]] · [[Shivon Zilis]] · [[PX 251]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-926", "exhibit": "DX 926", "party": "Defense", "type": "Text messages", "admitted_trial_day": "Day 8 (May 6, 2026)", "uploaded_box_pt": "2026-05-06T14:57:00", "uploader": "Morrison Foerster", "pages": 2, "size_bytes": 98323, "source_pdf": "DX-926.pdf", "pdf_url": "https://media.mts-in.com/DX-926.pdf", "body_markdown": "# DX-926 — Oct 19, 2020 Zilis-to-Altman: pinging Musk and the \"you should have gone with Tesla\" card\n\n> Shivon Zilis advising Sam Altman to keep Musk briefed: \"ping more often than less … as long as it's not feeling adversarial,\" and warning that Musk might pull \"the 'you should have gone with Tesla' card.\"\n\n## Document type\n**Text messages** (Short Message Report extract — 4-participant chat). One conversation, 3 messages on 10/19/2020, four participants visible: Sam Altman (two endpoints), Shivon Zilis, and an additional redacted contact. Bates OPENAI_MUSK00018213–00018214.\n\n## Logistics\n- **Trial admission:** Day 8 (May 6, 2026) — Murati live; Zilis live; Toner video begins.\n- **Box upload:** 2026-05-06 14:57:00 PT — Day-8 batch.\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~98 KB, 2 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `0926.pdf`.\n\n## Transcribed text\n\n> **Short Message Report**\n> Conversations: 1 · Total Messages: 3 · Date Range: 10/19/2020\n> Participants: Sam Altman, Sam Altman, Shivon Zilis, [redacted]\n>\n> **shivon zilis** [10/19/2020, 12:00 AM]: Yeah tbh I would recommend pinging more often than less on stuff like this as long as it's not feeling adversarial. If he does give you advice that helps, amazing. Even if not, if he's cordial he'll be less inclined to surprise you on Twitter. And if he is rough, it will be the same rough you'll see on Twitter so at least you'll know\n>\n> **shivon zilis** [10/19/2020, 12:02 AM]: Low probability, but the only thing I wonder is if he'll pull the **\"you should have gone with Tesla\"** card on you, so maybe just think through how you'd want to productively answer that one if it comes up.\n>\n> **Sam Altman** [10/19/2020, 12:05 AM]: ok! thanks\n\n## Commentary\n\nDX 926 is a small but pointed defense exhibit dated within weeks of Musk's **September 24, 2020 \"captured by Microsoft\"** tweet ([[Key Themes]] SOL section) — and it captures Zilis (who worked for Musk) coaching Altman on Musk-management strategy. Two beats matter on [[Day 8|Day 8]] (Zilis live): (i) Zilis identifies Musk's recurring \"**you should have gone with Tesla**\" gambit a year and a half after [[DX-844]] (\"attempt that through Tesla instead\") and [[DX-761]] (\"move three or four people from OpenAI to Tesla\") — corroborating the defense's parallel-AGI-play framing; (ii) the \"ping more often than less\" advice supports the defense's statute-of-limitations theory that Musk was being kept abreast in 2020 and chose Twitter over a lawsuit. Altman's polite \"ok! thanks\" is consistent with Musk's own testimony that Altman \"immediately reached out to reassure me that OpenAI was on its — was staying on mission as a nonprofit\" (`042926TT.txt:2194`).\n\n---\n\n*See also:* [[Day 8|Day 8 digest]] · [[Shivon Zilis]] · [[Sam Altman]] · [[DX-761]] · [[DX-844]] · [[Key Themes]]\n"}
{"exhibit_id": "DX-927", "exhibit": "DX 927", "party": "Defense", "type": "Text messages", "admitted_trial_day": "Day 3 (April 29, 2026)", "uploaded_box_pt": "2026-04-29T15:37:12", "uploader": "Morrison Foerster", "pages": 2, "size_bytes": 85517, "source_pdf": "DX-927.pdf", "pdf_url": "https://media.mts-in.com/DX-927.pdf", "body_markdown": "# DX 927 — Oct 28, 2020 Altman → Musk \"advice on next Microsoft investment\" texts\n\n> A two-message text exchange on October 28, 2020 in which Sam Altman tells Elon Musk \"**i would love to get you some advice from you on the next microsoft investment we are thinking of**\" and Musk replies \"**Ok. I can talk probably tomorrow or the next day.**\"\n\n## Document type\n**Text messages, Cellebrite-style \"Short Message Report\" extraction.** Two pages: cover sheet (1 conversation, 2 messages, 5 participants — including two Elon Musk handles, one of which is `<erm@tesla.com>`, plus Sam Altman, with three handles fully redacted) plus one page of message-by-message transcript with timestamps in GMT-07:00. Bates 2024MUSK-0006317–6318. Document ID DX-0927.\n\n## Logistics\n- **Trial admission:** Day 3 (April 29, 2026). Used by defense during Musk cross to evidence ongoing Musk-Altman Microsoft consultation in late 2020 — relevant to the [[Statute of Limitations]] theory and to rebut the \"phase three\" sudden-disillusionment narrative.\n- **Box upload:** 2026-04-29 15:37:12 PT — Day 3 late-afternoon batch (clustered with DX 679, DX 827, DX 862; same-second timestamp as DX 862).\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm).\n- **File size:** ~83 KB, 2 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `927 .pdf` (note trailing space in filename).\n\n## Transcribed text\n\n> **Conversation:** [chat — handle redacted]\n> **Date Range:** 10/28/2020\n> **Participants:** Elon Musk <[redacted]>, Elon Musk <erm@tesla.com>, Elon Musk <[redacted]>, Sam Altman <[redacted]>, [one other redacted]\n>\n> **Messages in chronological order (GMT -07:00):**\n>\n> **SA** (Sam Altman) — 10/28/2020, 11:37 AM\n> i would love to get you some advice from you on the next microsoft investment we are thinking of if you have time in the next week or so\n>\n> Receipts • Elon Musk [redacted] [R: 10/28/2020, 2:52 PM]\n>\n> **EM** (Elon Musk) — 10/28/2020, 2:53 PM\n> Ok. I can talk probably tomorrow or the next day.\n>\n> Receipts • Sam Altman [redacted] [D: 10/28/2020, 2:53 PM]\n\n## Commentary\n\nDX 927 is small but useful for the defense limitations narrative. By October 28, 2020 — five weeks after the [[PX 1504]] Microsoft \"exclusive license GPT-3\" blog post — Altman is openly soliciting Musk's \"**advice**\" on the next Microsoft investment, and Musk is agreeing to talk. Defense uses this two-line exchange to argue: (i) Musk continued to be treated by OpenAI's CEO as a sounding-board on Microsoft economics in late 2020, the same period plaintiffs' [[PX 103]] theory says was a § 17510.6 \"solicitation\" period — which the defense reads as further evidence Musk had actual contemporaneous knowledge of the OpenAI-Microsoft commercial relationship; (ii) the absence of any Musk objection in this thread (or any followed-on objection in the record from this period) cuts against the \"phase three\" 2022–2023 sudden-disillusionment frame plaintiffs presented in their direct case (see [[Key Themes]] § \"Phase three\"); and (iii) by 2020 Musk was using `erm@tesla.com` for these texts, a small but useful indicator of where Musk was concentrating his attention by then. Cross-reference: [[PX 1504]] (Sept 22, 2020 Microsoft blog post); [[PX 103]] (July 2020 Clark→Birchall rent reimbursement); [[PX 296]] (Oct 2022 \"bait and switch\") — DX 927 sits between the latter two on the timeline.\n\n---\n*See also:* [[Day 3|Day 3 digest]] · [[Sam Altman]] · [[PX 1504]] · [[PX 103]] · [[PX 296]] · [[Statute of Limitations]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-100", "exhibit": "PX 100", "party": "Plaintiffs", "type": "Letter", "admitted_trial_day": "Day 4 (April 30, 2026)", "uploaded_box_pt": "2026-04-30T15:26:46", "uploader": "Someone", "pages": 1, "size_bytes": 271979, "source_pdf": "PX-100.pdf", "bates": "2024Excession-0000002", "pdf_url": "https://media.mts-in.com/PX-100.pdf", "body_markdown": "# PX 100 — OpenAI thank-you letter for Musk's four-Tesla Model 3 donation (January 24, 2018)\n\n> Chris Clark's January 24, 2018 letter on OpenAI letterhead acknowledging Musk's donation of four 2017 Tesla Model 3 sedans (with VINs) on October 2, 2017, plus 2018 upgrades — total combined valuation $262,399.50.\n\n## Document type\n**Letter.** One-page tax-acknowledgement letter on OpenAI letterhead (3180 18th Street, Suite 100, San Francisco, CA 94110), dated January 24, 2018, signed by Chris Clark, Chief Operating Officer, OpenAI, Inc. (chris@openai.com). Includes a four-row vehicle table with VINs and per-vehicle valuations. Marked \"Confidential.\" Bates 2024Excession-0000002.\n\n## Logistics\n- **Trial admission:** Day 4 (April 30, 2026) — Musk cross + redirect; Birchall direct + cross. One of the in-kind contribution receipts in plaintiffs' contribution-tracking case.\n- **Box upload:** 2026-04-30 15:26:46 PT — uploaded fifteen seconds before PX 17 in the same Day 4 cross-prep batch as PX 86 (the YC.org receipt for the same donor).\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~272 KB, 1 page.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `100.pdf`.\n\n## Transcribed text\n\n> **OpenAI**\n> 3180 18th Street, Suite 100\n> San Francisco, CA 94110\n>\n> January 24, 2018\n>\n> Mr. Elon Musk\n> c/o CTC myCFO\n> PO Box 10195, Dept 863\n> Palo Alto, CA 94303\n>\n> Dear Musk:\n>\n> This letter is to gratefully acknowledge, for your tax records, our receipt of your generous gift of the four vehicles listed below on October 2, 2017 with a total value of $248,295.00.\n>\n> We also received additional upgrades to one of the vehicles on January 24, 2018 valued at 14,104.50.\n>\n> As summarized below, the total combined donation valuation is $262,399.50:\n>\n> | Vehicle Description | VIN | 2017 Donation Value | 2018 Additional Upgrades Donation Value | Total Combined Donation Value |\n> |---|---|---|---|---|\n> | 2017 Tesla Model 3 4-Door Sedan, Mileage: 50 | 5YJ3E1EA7HF000127 | $66,144.00 | $0 | $66,144.00 |\n> | 2017 Tesla Model 3 4-Door Sedan, Mileage: 50 | 5YJ3E1EA3HF000125 | $66,144.00 | $0 | $66,144.00 |\n> | 2017 Tesla Model 3 4-Door Sedan, Mileage: 50 | 5YJ3E1EA0HF000129 | $66,144.00 | $0 | $66,144.00 |\n> | 2017 Tesla Model 3 4-Door Sedan, Mileage: 50 | 5YJ3E1EA3HF000187 | $49,863.00 | $14,104.50 | $63,967.50 |\n> | **TOTAL** | | **$248,295.00** | **$14,104.50** | **$262,399.50** |\n>\n> No goods or services were provided to you in exchange for this gift. Therefore, the full amount of your gift qualifies as a charitable contribution for federal tax purposes. For reference, OpenAI's EIN is: 81-0861541.\n>\n> Please let me know if we can provide any additional information, and thank you again for your generous gift!\n>\n> Sincerely,\n>\n> *(signature)*\n>\n> Chris Clark\n> Chief Operating Officer\n> OpenAI, Inc.\n> chris@openai.com\n\n## Commentary\n\nPX 100 is a non-cash contribution receipt — the sort of in-kind line item Birchall walked the jury through on Day 4 (PX 112A: ~60 contributions totaling $38,191,066). Four 2017 Tesla Model 3s, all with 50 miles on the odometer, donated by Musk to OpenAI on October 2, 2017 — i.e., a few weeks after the [[PX 157|\"final straw\" / \"Honest Thoughts\" email]] of September 20, 2017 in which Musk had told Sutskever and Brockman \"I will no longer fund OpenAI until you have made a firm commitment to stay.\" That timing is significant for both sides: plaintiffs use it as evidence Musk *did* keep contributing into late 2017 even after the \"final straw\" rupture, undercutting the defense's \"picked up his marbles\" framing; defense uses it to anchor the notion that the relationship was still functional in October 2017 and that the rupture was about \"unequivocal control,\" not the nonprofit form. The same Chris Clark who signed the 2017 [[PX 86|YC.org]] gift letter signs this one — by January 2018 he had moved into OpenAI directly as COO, the trajectory of which (and his July 2020 Pioneer-Building rent email) figures into plaintiffs' alternative § 17510.6 charitable-solicitation theory.\n\n---\n*See also:* [[Day 4|Day 4 digest]] · [[PX 17]] · [[PX 86]] · [[PX 157]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-103", "exhibit": "PX 103", "party": "Plaintiffs", "type": "Email", "admitted_trial_day": "Day 4 (April 30, 2026)", "uploaded_box_pt": "2026-04-30T15:26:46", "uploader": "Someone", "pages": 1, "size_bytes": 85420, "source_pdf": "PX-103.pdf", "pdf_url": "https://media.mts-in.com/PX-103.pdf", "body_markdown": "# PX 103 — July 22, 2020 Chris Clark → Jared Birchall, OpenAI rent reimbursement\n\n> Chris Clark's \"OpenAI Update\" email proposing that the OpenAI for-profit cover 100% of rent going forward and asking Birchall (the Musk Foundation/family office) for a final one-time donation toward landlord pass-throughs and security costs — the email plaintiffs call the **2020 charitable solicitation** and treat as their answer to the statute-of-limitations defense.\n\n## Document type\n**Email, plain text, single message (forwarded).** A short email from Chris Clark <chris@openai.com> to Jared Birchall <jared@excession.com>, dated Wed, Jul 22, 2020 at 6:48 PM, subject \"OpenAI Update.\" Sender's signature partially redacted. Bates EXMF-0003682.\n\n## Logistics\n- **Trial admission:** Day 4 (April 30, 2026). Per [[Key Themes]]: \"PX 103 — July 2020 Clark email re: rent reimbursement,\" used by plaintiffs as the alleged 2020 \"solicitation\" that would defeat the statute of limitations on the § 17510.6 charitable-solicitation theory tied to the Pioneer Building.\n- **Box upload:** 2026-04-30 15:26:46 PT — late-afternoon Day 4 batch.\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~83 KB, 1 page.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `103.pdf`.\n\n## Transcribed text\n\n> ---------- Forwarded message ---------\n> **From:** Chris Clark <chris@openai.com>\n> **Date:** Wed, Jul 22, 2020 at 6:48 PM\n> **Subject:** OpenAI Update\n> **To:** Jared Birchall <jared@excession.com>\n>\n> Jared,\n>\n> Thanks for your call last week. I've confirmed with the team that OpenAI's for-profit entity is happy to cover 100% of the rent effective immediately.\n>\n> Since the non-profit needs to remain on the lease, it would greatly help the nonprofit org if you're willing to assist with covering a portion of the anticipated near term landlord passthroughs and security costs.\n>\n> With rent out of the picture, a final one-time donation might look like:\n>\n> - 50% of the $780k anticipated landlord project passthrough ($390k)\n> - 50% of the $20k per month 24/7 security costs over the next 18 months ($180k)\n> - Total: $570k\n>\n> Alternatively, you could simply keep the current $290k monthly Fidelity grants going through the end of September and get to roughly the same amount.\n>\n> We certainly understand if you'd prefer to just stop everything now. We are extremely grateful for all you've done to support OpenAI to date, and we want you all to do whatever you feel is most fair.\n>\n> Feel free to call any time if you want to discuss.\n>\n> Thanks again,\n> Chris\n\n## Commentary\n\nPX 103 is the linchpin of plaintiffs' alternative § 17510.6 charitable-solicitation theory. The statute-of-limitations math for that count requires an act of solicitation within the limitations period — and this July 2020 ask, in plaintiffs' framing, is exactly that: a request to the Musk side to \"assist with covering a portion of the anticipated near term landlord passthroughs and security costs,\" coupled with a reminder of the existing **$290k monthly Fidelity grants** (i.e., grants out of a Musk-related donor-advised fund). Note also Clark's revealing structural concession: \"Since the **non-profit needs to remain on the lease**\" — language plaintiffs use to argue that even in 2020 OpenAI's nonprofit shell was being kept in place specifically to anchor obligations the for-profit could not assume on its own. The defense reading on cross is that \"**OpenAI's for-profit entity is happy to cover 100% of the rent effective immediately**\" is the opposite of a continuing solicitation: the for-profit was offering to take over the lease. Cross-reference: the Pioneer Building thread, including [[DX 539]] (Musk: \"I don't want Sam on the lease,\" June 2016), Birchall's January 2017 note to Sam Teller that \"Elon holds 100 percent of the power and authority regarding the building\" (DX 600, not in this chunk), and the January 2019 reimbursement negotiation (DX 857, not in this chunk).\n\n---\n*See also:* [[Day 4|Day 4 digest]] · [[Pioneer Building]] · [[Statute of Limitations]] · [[DX 539]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-105", "exhibit": "PX 105", "party": "Plaintiffs", "type": "Text messages", "admitted_trial_day": "Day 3 (April 29, 2026)", "uploaded_box_pt": "2026-04-29T15:09:19", "uploader": "Someone", "pages": 3, "size_bytes": 191182, "source_pdf": "PX-105.pdf", "pdf_url": "https://media.mts-in.com/PX-105.pdf", "body_markdown": "# PX 105 — Oct 1, 2020 Musk ↔ Altman \"appear hypocritical / change the name\" texts\n\n> A 14-message text thread on October 1, 2020 in which Musk tells Altman OpenAI shouldn't \"be (or at least appear to be) hypocritical\" and writes \"**At least change the name**\" — Altman replies \"agreed we should try not to appear hypocritical.\"\n\n## Document type\n**Text messages, Cellebrite-style \"Short Message Report\" extraction.** Three pages: cover sheet (1 conversation, 14 messages, 5 participants — including two Elon Musk handles, one shown as `erm@tesla.com`, plus Sam Altman, all other handles redacted) and two pages of message-by-message transcript with timestamps in GMT-07:00. Bates 2024MUSK-0006301–6303.\n\n## Logistics\n- **Trial admission:** Day 3 (April 29, 2026). No prior wiki reference; admission day inferred from Box upload date and Day 3 mid-afternoon plaintiffs' batch (clustered with PX 158, PX 236).\n- **Box upload:** 2026-04-29 15:09:19 PT.\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~187 KB, 3 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `105.pdf`.\n\n## Transcribed text\n\n> **Conversation:** [chat — handle redacted]\n> **Date Range:** 10/1/2020\n> **Participants:** Elon Musk [redacted], Elon Musk [redacted], Elon Musk <erm@tesla.com>, Sam Altman [redacted]\n>\n> **Messages in chronological order (GMT -07:00):**\n>\n> **SA** (Sam Altman) — 10/1/2020, 11:02 AM\n> Saw your feedback on Twitter last week. I'm happy to talk about this if you'd like, but I think there's no way we can hold a candle to DeepMind without many billions of dollars, and MSFT still seems like the best way for us to get that with the least compromise. We gave MSFT a copy of GPT-3 to use in their own products, but we still get to retain autonomy to release our work ourselves (e.g., we can and will continue to provide API access to the most powerful language model in existence to everyone). If you've got any feedback about how to do better, I'd love to hear it. Thanks! Sam.\n>\n> **EM** (Elon Musk) — 10/1/2020, 11:08 AM\n> Yeah, we should talk\n>\n> **EM** — 10/1/2020, 11:09 AM\n> I don't think it's a winning approach to be (or at least appear to be) hypocritical\n>\n> **EM** — 10/1/2020, 11:09 AM\n> At least change the name\n>\n> **SA** — 10/1/2020, 11:09 AM\n> happy to talk whenever you're free--agreed we should try not to appear hypocritical\n>\n> **EM** — 10/1/2020, 11:23 AM\n> Sounds good. I'm in Texas and about to do a series of Starship engineering reviews, but free tonight.\n>\n> **SA** — 10/1/2020, 6:40 PM\n> im free for the rest of the night\n>\n> **SA** — 10/1/2020, 10:21 PM\n> have a good flight--super open to feedback about how to balance openness and safety, and how to talk about it\n>\n> **EM** — 10/1/2020, 10:34 PM\n> Liked \"have a good flight--super open to feedback about how to balance openness and safety, and how to talk about it\"\n>\n> **EM** — 10/1/2020, 10:34 PM\n> Sigh\n>\n> **EM** — 10/1/2020, 10:34 PM\n> https://twitter.com/interviewopen/status/1311902233148055552?s=10\n>\n> **EM** — 10/1/2020, 10:34 PM\n> [Attachment: ~_Library_SMS_Attachments_a2_02_1EB4A92F-88C7-43E8-9BED-5F23FACAAEFC_475B254C-B4D2-4A9F-A903-8E2226D36140.pluginPayloadAttachment (6 KB)]\n>\n> **EM** — 10/1/2020, 10:34 PM\n> 475B254C-B4D2-4A9F-A903-8E2226D36140.pluginPayloadAttachment\n>\n> **SA** — 10/1/2020, 10:35 PM\n> [smiley emoji]\n>\n> **SA** — 10/1/2020, 10:39 PM\n> we finally just got a full time PR person (steve dowling who previously ran it at apple) so i am hopeful we can start getting PR right…\n\n## Commentary\n\nPX 105 is plaintiffs' anchor for the 2020 inflection point in Musk's disillusionment narrative — one week after the September 2020 GPT-3-to-Microsoft exclusive license (which Musk publicly criticized on Twitter). The \"**At least change the name**\" line is a 2020 echo of Musk's [[Key Themes#The \"OPEN\" in OpenAI|\"OPEN\" means open source]] trial testimony, used by plaintiffs as contemporaneous evidence that Musk was already complaining the name had become misleading. Altman's \"we should try not to appear hypocritical\" — agreement, not deflection — is what plaintiffs lean on to argue OpenAI itself recognized the conflict. The exhibit also charts the [[Microsoft thread]]: Altman's \"MSFT still seems like the best way for us to get that with the least compromise\" is the defense's \"Musk-knew-and-acquiesced\" beat (he did not sue in 2020). Note that Musk is texting from `erm@tesla.com` here — see [[DX 748]] / [[DX 749]] companion data on Musk-Tesla integration.\n\n---\n*See also:* [[Day 3|Day 3 digest]] · [[Sam Altman]] · [[PX 233]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-112", "exhibit": "PX 112", "party": "Plaintiffs", "type": "Demonstrative", "admitted_trial_day": "Day 4 (April 30, 2026)", "uploaded_box_pt": "2026-04-30T15:26:47", "uploader": "Someone", "pages": 4, "size_bytes": 333501, "source_pdf": "PX-112.pdf", "pdf_url": "https://media.mts-in.com/PX-112.pdf", "body_markdown": "# PX 112 — \"Musk's Contributions\" demonstrative ($38,191,066 total)\n\n> Plaintiffs' four-page demonstrative table itemizing every Musk contribution to OpenAI from May 27, 2016 through September 14, 2020, totaling **$38,191,066** — used by Birchall on direct on Day 4 and the source of the trial's most-repeated number.\n\n## Document type\n**Demonstrative — chart/table summary exhibit.** Tabular slide deck listing approximate date, amount, source, immediate payer, immediate recipient, ultimate recipient/beneficiary, and supporting underlying exhibit. Cited as **PX 112A** in [[Key Themes]].\n\n## Logistics\n- **Trial admission:** Day 4 (April 30, 2026) — Birchall direct: \"Birchall confirms the precise number on direct: **$38,191,066** in ~60 contributions (PX 112A). `(043026TT.txt:4291)`\" (See [[Key Themes]] §\"The $1B / $38M gap.\")\n- **Box upload:** 2026-04-30 15:26:47 PT.\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~326 KB, 4 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `112.pdf`.\n\n## Transcribed text\n\n> **MUSK'S CONTRIBUTIONS**\n>\n> | Approximate Date | Amount | Source | Immediate Payer | Immediate Recipient | Ultimate Recipient/Beneficiary | Exhibit |\n> |---|---|---|---|---|---|---|\n> | 5/27/2016 | $500,000 | Elon Musk | Elon Musk | YC Org | OpenAI, Inc. | PX73 |\n> | 6/2/2016 | $5,000,000 | Elon Musk | Vanguard DAF | YC Org | OpenAI, Inc. | PX60 (-241) |\n> | 6/15/2016 | $141,667 | Musk Industries LLC | Musk Industries LLC | Bridgeton Pioneer | OpenAI, Inc. | PX75; PX76 |\n> | 8/19/2016 | $4,500,000 | Musk Foundation | Vanguard DAF | YC Org | OpenAI, Inc. | PX60 (-281) |\n> | 9/22/2016 | $142,000 | Musk Foundation | Vanguard DAF | YC Org | OpenAI, Inc. | PX60 (-285) |\n> | 10/17/2016 | $142,000 | Musk Foundation | Vanguard DAF | YC Org | OpenAI, Inc. | PX60 (-289) |\n> | 11/11/2016 | $750,000 | Musk Foundation | Vanguard DAF | YC Org | OpenAI, Inc. | PX60 (-293) |\n> | 11/15/2016 | $142,000 | Musk Foundation | Vanguard DAF | YC Org | OpenAI, Inc. | PX60 (-297) |\n> | 11/30/2016 | $4,250,000 | Musk Foundation | Vanguard DAF | YC Org | OpenAI, Inc. | PX60 (-301) |\n> | 12/15/2016 | $142,000 | Musk Foundation | Vanguard DAF | YC Org | OpenAI, Inc. | PX60 (-305) |\n> | 1/17/2017 | $142,000 | Musk Foundation | Vanguard DAF | YC Org | OpenAI, Inc. | PX60 (-269) |\n> | 2/15/2017 | $142,000 | Musk Foundation | Vanguard DAF | YC Org | OpenAI, Inc. | PX60 (-273) |\n> | 2/27/2017 | $5,000,000 | Musk Foundation | Musk Foundation | YC Org | OpenAI, Inc. | PX86 |\n> | 3/16/2017 | $175,000 | Musk Foundation | Vanguard DAF | YC Org | OpenAI, Inc. | PX60 (-277) |\n> | 4/18/2017 | $175,000 | Musk Foundation | Vanguard DAF | OpenAI, Inc. | OpenAI, Inc. | PX60 (-253) |\n> | 5/15/2017 | $175,000 | Musk Foundation | Vanguard DAF | OpenAI, Inc. | OpenAI, Inc. | PX60 (-257) |\n> | 5/26/2017 | $5,000,000 | Musk Foundation | Musk Foundation | YC Org | OpenAI, Inc. | PX87 |\n> | 6/15/2017 | $175,000 | Musk Foundation | Vanguard DAF | OpenAI, Inc. | OpenAI, Inc. | PX60 (-261) |\n> | 7/18/2017 | $175,000 | Musk Foundation | Fidelity DAF | OpenAI, Inc. | OpenAI, Inc. | PX91 (-761) |\n> | 7/20/2017 | $250,000 | Musk Foundation | Fidelity DAF | YC Org | OpenAI, Inc. | PX92 |\n> | 8/14/2017 | $175,000 | Musk Foundation | Fidelity DAF | OpenAI, Inc. | OpenAI, Inc. | PX91 (-763) |\n> | 9/15/2017 | $175,000 | Musk Foundation | Fidelity DAF | OpenAI, Inc. | OpenAI, Inc. | PX91 (-765) |\n> | 9/29/2017 | $85,000 | Musk Foundation | Fidelity DAF | OpenAI, Inc. | OpenAI, Inc. | PX91 (-767) |\n> | 10/2/2017 | $248,295 | Elon Musk | Elon Musk | OpenAI, Inc. | OpenAI, Inc. | PX100 |\n> | 10/16/2017 | $235,000 | Musk Foundation | Fidelity DAF | OpenAI, Inc. | OpenAI, Inc. | PX91 (-769) |\n> | 11/14/2017 | $235,000 | Musk Foundation | Fidelity DAF | OpenAI, Inc. | OpenAI, Inc. | PX91 (-771) |\n> | 12/14/2017 | $235,000 | Musk Foundation | Fidelity DAF | OpenAI, Inc. | OpenAI, Inc. | PX91 (-773) |\n> | 1/18/2018 | $290,000 | Musk Foundation | Fidelity DAF | OpenAI, Inc. | OpenAI, Inc. | PX91 (-775) |\n> | 1/24/2018 | $14,105 | Elon Musk | Elon Musk | OpenAI, Inc. | OpenAI, Inc. | PX100 |\n> | 2/20/2018 | $390,000 | Musk Foundation | Fidelity DAF | OpenAI, Inc. | OpenAI, Inc. | PX91 (-777) |\n> | 3/14/2018 | $290,000 | Musk Foundation | Fidelity DAF | OpenAI, Inc. | OpenAI, Inc. | PX91 (-779) |\n> | 4/16/2018 | $290,000 | Musk Foundation | Fidelity DAF | OpenAI, Inc. | OpenAI, Inc. | PX91 (-781) |\n> | 5/15/2018 | $290,000 | Musk Foundation | Fidelity DAF | OpenAI, Inc. | OpenAI, Inc. | PX91 (-783) |\n> | 6/14/2018 through 9/14/2020 | $290,000 (29 monthly entries) | Musk Foundation | Fidelity DAF | OpenAI, Inc. | OpenAI, Inc. | PX91 (-785 through -841) |\n>\n> **Total: $38,191,066**\n\n## Commentary\n\nPX 112 is plaintiffs' answer to defense's $1B-pledge-vs-$38M-actual-contribution attack — and is also the table on which that defense attack rests. The exhibit shows that **after September 2020 (the \"captured by Microsoft\" tweet), Musk's $290K monthly auto-contributions stopped**, and that the donation cadence shifted from corporate ($5M lump sums in 2016–17) to monthly DAF distributions of $290K once OpenAI moved to its for-profit structure. The chart also exposes the **donor-advised-fund plumbing** that powers plaintiffs' standing problem: most contributions flowed Musk Foundation → Vanguard/Fidelity DAF → OpenAI, meaning Musk irrevocably parted with the funds at the DAF step and could only \"advise\" their direction (see [[Key Themes]] §\"The 'donor-advised funds' / tax-deduction wrinkle\"). Birchall on cross conceded each contribution was tax-deductible. The total figure — $38,191,066 — is the number Musk on cross conceded (\"In strict monetary terms, I contributed 38 million\") and the foundation of every defense reference to \"less than 4 percent of what he promised.\" Cross-reference [[PX-91]] (Fidelity DAF distribution records).\n\n---\n*See also:* [[Day 4|Day 4 digest]] · [[Jared Birchall]] · [[PX-91]] · [[Key Themes]] · [[Elon Musk]]\n"}
{"exhibit_id": "PX-12", "exhibit": "PX 12", "party": "Plaintiffs", "type": "Email thread", "admitted_trial_day": "Day 2 (April 28, 2026)", "uploaded_box_pt": "2026-04-28T14:51:59", "uploader": "Someone", "pages": 3, "size_bytes": 692481, "source_pdf": "PX-12.pdf", "pdf_url": "https://media.mts-in.com/PX-12.pdf", "body_markdown": "# PX 12 — Nov 14–23, 2015 Musk ↔ Altman \"Thought Summary\" / naming-and-structure email thread\n\n> The earliest end-to-end exchange in evidence between Elon Musk and Sam Altman setting up what would become OpenAI — a three-page email chain in which they agree on an \"independent, pure play 501c3,\" debate names (\"Freemind,\" \"Axon\"), and discuss Musk's involvement (chair vs. co-CEO) and the cash-bonus comp model with optional conversion to SpaceX/YC stock.\n\n## Document type\n**Email thread, plain text, four messages (Nov 14, 22, 23, 2015).** Subject lines: \"Re: Great Acton quotes\" and \"Re: Thought Summary.\" Bates 2024MUSK-0010041–0010043. Stamped \"Highly Confidential-AEO.\" Sender names of Sam Altman are visibly redacted in black bars on each header (forwarded via a redirect address `on behalf of Sam Altman <sam@ycombinator.com>`).\n\n## Logistics\n- **Trial admission:** Day 2 (April 28, 2026) — used during Musk direct examination, in the opening sequence laying out the founding-of-OpenAI narrative (the \"I started a nonprofit\" framing — see [[Key Themes]] § \"Stole a charity\").\n- **Box upload:** 2026-04-28 14:51:59 PT — Day 2 mid-afternoon batch.\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~676 KB, 3 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `012.pdf`.\n\n## Transcribed text\n\n> **From:** \"Elon Musk\" <erm@spacex.com>\n> **To:** \"Sam Altman\" <sam@ycombinator.com>\n> **Subject:** Re: Thought Summary\n> **Date:** Mon, 23 Nov 2015 03:50:06 -0000\n>\n> Great.\n>\n> Something Turing-related that doesn't sound too ominous might be good. Want to avoid the Turing Test association though, as that sounds too much like we are replacing humans.\n>\n> ---\n>\n> **From:** [Sam Altman] on behalf of Sam Altman <sam@ycombinator.com>\n> **Sent:** Sunday, November 22, 2015 7:36 PM\n> **To:** Elon Musk\n> **Subject:** Re: Thought Summary\n>\n> Ok, let me work on a draft of how to step up the bonuses for the senior positions.\n>\n> I'm warming up fast to freemind. It definitely coveys the right spirit. Also thinking about names related to Turing somehow.\n>\n> I think both as co-chairs works best. We can evolve from there as needed.\n>\n> ---\n>\n> On Sun, Nov 22, 2015 at 7:32 PM, Elon Musk <erm@spacex.com> wrote:\n>\n> Great, sounds like we are independently following the same thought tracks.\n>\n> No problem with YC support in the beginning. That's normal for a lot of companies, so I don't think it materially affects the neutrality position.\n>\n> The \"mind\" part of Freemind is partly an intentional philosophical counter to Deepmind and also, as mentioned, I couldn't find a synonym for \"mind\" that wasn't terrible.\n>\n> I like Axon as an option, although it does slightly sound like Google Brain or, more generally, that we think digital intelligence consists of brain emulation. Pretty much all names suck in the beginning though. Given several choices, I tend to favor names that convey the mission of the company and hopefully have a positive impact on recruitment.\n>\n> Yeah, my role should be clear and I have to bite the bullet on admitting real involvement. This will come as a shocker to many, but so be it. Can't be lukewarm about this.\n>\n> In terms of title, what about me as chair and you as CEO or us both as co-chairs? My main concern about being called a co-CEO, even if that is de facto true, is that it might cause problems at SpaceX and Tesla, where they know I'm already overbooked.\n>\n> ---\n>\n> **From:** [Sam Altman] on behalf of Sam Altman <sam@ycombinator.com>\n> **Sent:** Sunday, November 22, 2015 7:08 PM\n> **To:** Elon Musk\n> **Subject:** Re: Thought Summary\n>\n> Yes — I think an independent 501c3 is correct. Are you ok if, initially, it shares a building with YC and YC helps with legal stuff etc? I think that will save us a few months while we hire dedicated people.\n>\n> I like the idea of defaulting to a cash bonus that stacks each year — I'd independently come to that same conclusion after talking to Greg yesterday. Offering them the option to convert it to SpaceX or YC stock sounds great to me.\n>\n> Making Tesla data available would be incredible. I think most YC companies will offer to do the same.\n>\n> Freemind seems pretty good to me but I'd to keep thinking — it may be a little too close to Deepmind. I really like the word free here though — freethink? freemerge?\n>\n> Best idea I've had so far is Axon, which I think we could get .com for from TPG-Axon. Intelligence.com looks getable, but perhaps too generic.\n>\n> I'll keep thinking.\n>\n> Half a day per week sounds awesome. I assume that if it gets to the point where AGI looks like it's about to happen, you and I will both find a way to spend a lot of time on it. And if it seems further off, then I expect you'll be very busy getting to Mars.\n>\n> Have you given any thought to how you want to describe you involvement? I'm fine with anything from \"advisor\" to \"co-CEO\" but think it's probably good to have a clear answer.\n>\n> ---\n>\n> On Sun, Nov 22, 2015 at 6:54 PM, Elon Musk <erm@spacex.com> wrote:\n>\n> Had a great call with Greg. I'm super impressed with everyone so far. This is a great team.\n>\n> Doing this as an independent, pure play 501c3, but with a crystal clear focus on the positive advent of strong AI distributed widely to humanity does seem to make the most sense to me. Please correct me if we aren't on the same page. The company would still aim to bring in revenue in excess of costs at some point, but positive net revenue would just flow to cash reserves.\n>\n> On the comp front, I think having the default approach be a cash salary and a cash bonus that vests over four years and stacks with refresher vesting bonuses every year is the way to go. People could, at their option, if you are amenable, convert that cash amount to YC stock. I'm also fine if they want to convert some or all to SpaceX stock. I can pretty much do what I want on the SpaceX side, as it is private (thank goodness).\n>\n> Also, we will have insane amounts of real world sensor data — several orders of magnitude greater than any other company — at Tesla that the team could use when helpful.\n>\n> Was thinking a bit about names. It is always super hard to come up with one that doesn't suck and isn't already taken by someone that won't give it up. Least bad name I have thought of so far is Freemind. Conveys the sense that we are trying to create digital intelligence that will be freely available to all — the opposite of Deepmind's one-ring-to-rule-them-all approach. Having \"mind\" in there isn't great, but the synonyms suck and their name has been subsumed in the public sphere by Google/Alphabet. You come up with anything?\n>\n> On the time allocation front, I've decided to dedicate whatever amount of my time that is materially useful, even though it will come at some cost to SpaceX and Tesla. If I really believe that this is potentially the biggest near-term existential threat, then action should follow belief. That said, this probably amounts to half a day per week (higher in the beginning), with more time allocated when we hit a key decision point or crisis.\n>\n> ---\n>\n> **From:** [Sam Altman] on behalf of Sam Altman <sam@ycombinator.com>\n> **Sent:** Saturday, November 14, 2015 6:04 PM\n> **To:** Elon Musk\n> **Subject:** Re: Great Acton quotes\n>\n> the second is one of my all-time favorite quotes!\n>\n> https://www.youtube.com/watch?v=f-bwW1czuL4&feature=youtu.be&t=122\n>\n> ---\n>\n> On Sat, Nov 14, 2015 at 5:19 PM, Elon Musk <erm@spacex.com> wrote:\n>\n> \"Liberty consists in the division of power. Absolutism, in concentration of power.\"\n>\n> And\n>\n> **\"There is no error so monstrous that it fails to find defenders among the ablest men\"**\n\n## Commentary\n\nPX 12 is plaintiffs' founding-document exhibit — proof, in Musk's and Altman's own words, that the entity they were jointly creating in November 2015 was an **\"independent, pure play 501c3\"** with mission as the binding constraint and revenue as a residual, not a goal. Plaintiffs use it as the textual anchor for the [[Key Themes|\"stole a charity\"]] frame: the founders agreed in writing that \"positive net revenue would just flow to cash reserves.\" It also contains two soft notes the defense exploits: Musk's stated time commitment of \"half a day per week\" undercuts the later \"I funded everything\" framing, and the early willingness to allow comp conversion to \"SpaceX or YC stock\" complicates the pure-charity narrative — see Birchall's \"tax deduction every step of the way\" testimony in [[Key Themes]] § \"donor-advised funds.\" The Lord Acton epigraph that opens the chain (\"Liberty consists in the division of power. Absolutism, in concentration of power.\") becomes a defense talking point on cross — the [[PX-157|\"unequivocal control\"]] theme turns into the climax of: in 2017, Musk asked for exactly the kind of unilateral control he had just told Altman in 2015 was the definition of tyranny.\n\n---\n*See also:* [[Day 2|Day 2 digest]] · [[PX-157]] · [[Sam Altman]] · [[Elon Musk]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-1250-16", "exhibit": "PX 1250.16", "party": "Plaintiffs", "type": "Personal journal entry", "admitted_trial_day": "Day 7 (May 5, 2026)", "uploaded_box_pt": "2026-05-05T15:54:33", "uploader": "Morrison Foerster", "pages": 5, "size_bytes": 199920, "source_pdf": "PX-1250-16.pdf", "pdf_url": "https://media.mts-in.com/PX-1250-16.pdf", "body_markdown": "# PX 1250.16 — Brockman journal, Aug 18, 2017 (single Bates page of PX 154)\n\n> The five-page Bates-stamped slipsheet around a single August 18, 2017 Brockman journal page — \"**have been thinking hard about what control really means and whether we should do it. the answer is emerging: definitely not unilateral control. no person should have control over what we're creating. and what's fair is an equal split, and we can't agree to anything less than that.**\"\n\n## Document type\n**Personal journal entry, single-page extract.** A five-page packet: page 1 cover slip \"**File Name : 2017-08-18.txt**\", page 2 the substantive prose passage, pages 3–5 blank. Stream-of-consciousness journal Brockman has kept on his laptop since 2010 (see [[Brockman Journal]]). Bates OPENAI_MUSK00039287–39291. **This is a single-Bates-page extract of [[PX 151|PX 154]]** — the chunk metadata column flags it as such (\"Bates page of PX 154; uploaded by defense\").\n\n## Logistics\n- **Trial admission:** Day 7 (May 5, 2026 — Brockman defense direct + Mr. Molo recross + DX 1252; Wu deposition). Used as a discrete exhibit on the [[Brockman Journal|Brockman Journal Aug 18, 2017]] passage during the defense direct/recross sequence.\n- **Box upload:** 2026-05-05 15:54:33 PT — Day 7 mid-afternoon batch.\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm) — note that this Bates page of PX 154 was uploaded by *defense*, not plaintiffs, even though the column-A canon ID is in the PX series. The \"PX\" label reflects that the underlying document is a plaintiffs' exhibit (PX 154); the Box upload is defense's discrete extract.\n- **File size:** ~195 KB, 5 pages (1 substantive page + cover + blanks).\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `1250.16.pdf`.\n\n## Transcribed text\n\n> **File Name : 2017-08-18.txt**\n>\n> [Page 2 — substantive content, all lowercase as in source:]\n>\n> have been thinking hard about what control really means and whether we should do it. the answer is emerging: definitely not unilateral control. no person should have control over what we're creating. and what's fair is an equal split, and we can't agree to anything less than that.\n>\n> [Pages 3–5: blank.]\n\n## Commentary\n\nPX 1250.16 is the **single-page Bates extract** of the Aug 18, 2017 entry — the same passage that appears in the larger [[PX 151|PX 154]] journal compilation. The defense's tactical reason to walk it in as a *separate exhibit* on Day 7 is that the Aug 18 entry is the cleanest, shortest, most-quotable Brockman-as-principled-co-founder line in the entire journal, and is the textual lead-in to the [[Brockman Journal|\"morally bankrupt\"]] reading: the founders' red line, in Brockman's own contemporaneous handwriting, was **\"definitely not unilateral control … no person should have control over what we're creating\"** — i.e., directed at the [[PX 153|cap-table proposal]] in which Musk would hold 51.20%. Plaintiffs' read: the same entry says \"what's fair is an equal split\" — i.e., the founders also wanted *more* personal equity. The lower-case stream-of-consciousness prose is an authenticity tell consistent with Brockman's stand testimony that the journal is \"very chain of thought … stream of consciousness\" (see [[Brockman Journal]]). Quoted at trial with precise transcript timestamps in [[Brockman Journal|the Brockman Journal wiki]] (5/5/2026 Testimony @ ~09:34 PT).\n\n---\n*See also:* [[Brockman Journal]] · [[PX 151]] · [[Day 7|Day 7 digest]] · [[Greg Brockman]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-1250-52", "exhibit": "DX 1250.52", "party": "Defendants", "type": "Personal journal entry (Bates pages from PX 154 — Brockman journal)", "admitted_trial_day": "Day 7 (May 5, 2026)", "uploaded_box_pt": "2026-05-05T15:54:33", "uploader": "Morrison Foerster", "pages": 13, "size_bytes": 250604, "source_pdf": "PX-1250-52.pdf", "pdf_url": "https://media.mts-in.com/PX-1250-52.pdf", "body_markdown": "# DX 1250.52 — Brockman journal, November 5, 2017 (\"Real decision is fire elon\" / \"we seem converged on the 'fire elon' route\")\n\n> Defendant exhibit excerpting Greg Brockman's November 5, 2017 laptop journal entry — the two specific Bates pages of the larger Brockman journal corpus that defense counsel introduced on Day 7 redirect, **admitted over Mr. Molo's hearsay objection as a prior consistent statement under FRE 801(d)(1)(B)** to rebut the express charge of recent fabrication.\n\n## Document type\n**Personal journal entry (Bates pages from PX 154 — Brockman journal).** Two short isolated lines drawn from Brockman's contemporaneous laptop diary file `2017-11-05.txt`. Despite the TSV's PX-1250-52 short ID, the exhibit sticker in the PDF labels this as a **defendant's exhibit (DX-1250.52)**. The pages reproduced are mostly white space with the handful of substantive lines visible on pages 6 and 7. Bates OPENAI_MUSK00039457–00039469.\n\n## Logistics\n- **Trial admission:** Day 7 (May 5, 2026) — Brockman defense direct + Mr. Molo recross + DX 1252; Wu deposition. Used during defense redirect of Brockman.\n- **Box upload:** 2026-05-05 15:54:33 PT.\n- **Uploader:** Morrison Foerster (OpenAI startup-fund defense firm — same firm uploading defense exhibits in this window).\n- **File size:** ~245 KB, 13 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `1250.52.pdf`.\n\n## Transcribed text\n\n**[Page 1 — file label]** `File Name : 2017-11-05.txt`\n\n**[Pages 2–5 — blank]**\n\n**[Page 6 — Bates 462]**\n\n> - real decision is fire elon.\n\n**[Page 7 — Bates 463]**\n\n> - ok, so we seem converged on the \"fire elon\" route.\n\n**[Pages 8–13 — blank]**\n\n## Commentary\n\nDX 1250.52 is the textual core of the [[Key Themes|defense's recontextualization]] of the Brockman journal. On Day 7's redirect, **Mr. Gerrada introduced this exhibit (and the larger DX 1252) admitted over Mr. Molo's hearsay objection as a prior consistent statement under FRE 801(d)(1)(B) to rebut Molo's express charge of recent fabrication.** The two lines reproduced — \"**Real decision is fire elon**\" (Bates 462) and \"**So we seem converged on the 'fire elon' route**\" (Bates 463) — are the contemporaneous Nov 5, 2017 evidence that the \"**fire elon**\" decision was already in the room *before* Brockman's Nov 6, 2017 \"**It'd be wrong to steal the nonprofit from him. To convert to B Corp without him. They'd be pretty morally bankrupt**\" entry. Defense's reading: \"morally bankrupt\" referred to *voting to remove Musk from the OpenAI board and then creating a B Corp without him* — not to the for-profit conversion *as such*. **Plaintiffs' counter** (Mr. Molo on recross): \"**It's just an after-the-fact manufactured excuse for the very honest emotion that you expressed in the journal at the time**\" *(5/5/2026 Testimony @ ~14:08 PT)*. This exhibit is also the primary illustration of the [[Brockman Journal|journal's selective unblinding]]: the defense uploaded only the two pages with substantive content, leaving the rest of the day's text blank/redacted from this excerpt — a curatorial choice plaintiffs flagged on cross. See [[PX-154]] for the Sept 12, 2017 entry from the same journal volume; see [[Day 7|Day 7 digest]] for the live admission ruling and Mr. Molo's recross frame.\n\n---\n*See also:* [[Day 7|Day 7 digest]] · [[Day 6|Day 6 digest]] · [[PX-154]] · [[Brockman Journal]] · [[Greg Brockman]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-14", "exhibit": "PX 14", "party": "Plaintiffs", "type": "Email thread", "admitted_trial_day": "Day 2 (April 28, 2026)", "uploaded_box_pt": "2026-04-28T14:51:59", "uploader": "Someone", "pages": 2, "size_bytes": 490394, "source_pdf": "PX-14.pdf", "pdf_url": "https://media.mts-in.com/PX-14.pdf", "body_markdown": "# PX 14 — Dec 8, 2015 Musk/Altman draft of OpenAI launch announcement\n\n> The original \"OpenAI is a non-profit artificial intelligence research company\" opening paragraphs — drafted by Musk on Dec 8, 2015 and revised by Altman the same morning — that became the founding public statement of OpenAI.\n\n## Document type\n**Email thread, plain text, two messages.** Musk → Altman (sam@ycombinator.com), 9:29 AM PST, subject \"Draft opening paragraphs\"; Altman → Musk reply (\"how is this?\") at 10:34:59 AM with a tightened rewrite. Bates 2024MUSK-0000366–0000367.\n\n## Logistics\n- **Trial admission:** Day 2 (April 28, 2026) — openings + start of Musk direct. The exhibit is part of plaintiffs' founding-purpose chain that culminates in the public OpenAI launch the next day (Dec 11, 2015).\n- **Box upload:** 2026-04-28 14:51:59 PT — Day 2 mid-afternoon batch (clustered with PX 152 and other launch-window documents).\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~479 KB, 2 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `014.pdf`.\n\n## Transcribed text\n\n> **From:** Elon Musk\n> **To:** sam@ycombinator.com\n> **Subject:** Draft opening paragraphs\n> **Date:** Tuesday, December 8, 2015 9:29:15 AM\n>\n> It is super important to get the opening summary section right. This will be what everyone reads and what the press mostly quotes. The whole point of this release is to attract top talent. Not sure Greg totally gets that.\n>\n> ----\n>\n> OpenAI is a non-profit artificial intelligence research company with the goal of advancing digital intelligence in the way that is most likely to benefit humanity as a whole, unencumbered by an obligation to generate financial returns.\n>\n> The underlying philosophy of our company is to disseminate AI technology as broadly as possible as an extension of all individual human wills, ensuring, in the spirit of liberty, that the power of digital intelligence is not overly concentrated and evolves toward the future desired by the sum of humanity.\n>\n> The outcome of this venture is uncertain and the pay is low compared to what others will offer, but we believe the goal and the structure are right. We hope this is what matters most to the best in the field.\n\n> **From:** Sam Altman (on behalf of Sam Altman)\n> **To:** Elon Musk\n> **Subject:** Re: Draft opening paragraphs\n> **Date:** Tuesday, December 8, 2015 10:34:59 AM\n>\n> how is this?\n>\n> ---\n>\n> OpenAI is a non-profit artificial intelligence research company with the goal of advancing digital intelligence in the way that is most likely to benefit humanity as a whole, unencumbered by an obligation to generate financial returns.\n>\n> Because we don't have any financial obligations, we can focus on the maximal positive human impact and disseminating AI technology as broadly as possible. We believe AI should be an extension of individual human wills and, in the spirit of liberty, not be concentrated in the hands of the few.\n>\n> The outcome of this venture is uncertain and the pay is low compared to what others will offer, but we believe the goal and the structure are right. We hope this is what matters most to the best in the field.\n>\n> On Tue, Dec 8, 2015 at 9:29 AM, Elon Musk <erm@spacex.com> wrote: [Musk's draft, repeated above]\n\n## Commentary\n\nPX 14 is the foundational document for plaintiffs' \"stole a charity\" theory: the words *non-profit*, *unencumbered by an obligation to generate financial returns*, and *not be concentrated in the hands of the few* are all present in the **first draft**, three days before the launch — co-written by the two named defendants. Plaintiffs use it to anchor every later \"we are still on mission\" statement back to a written, contemporaneous, authored-by-Altman commitment about what OpenAI was and was not. The phrase **\"the pay is low compared to what others will offer\"** is also a standalone plaintiffs' point: Brockman's eventual ~$30 billion equity stake (see [[Key Themes]]) is hard to square with a recruiting pitch that explicitly told candidates to expect below-market pay. Defense will argue this is a marketing draft, not a legal undertaking — and that \"non-profit\" did not foreclose the later capped-profit subsidiary structure. Pair with [[PX 5]] (the Dec 11, 2015 launch blog) and [[PX 152]] (Aug 2017 Musk \"I've had enough\" reply) to see the same authors in the same email chain, two years apart.\n\n---\n*See also:* [[Day 2|Day 2 digest]] · [[PX 5]] · [[PX 152]] · [[Sam Altman]] · [[Elon Musk]] · [[Greg Brockman]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-15", "exhibit": "PX 15", "party": "Plaintiffs", "type": "Email", "admitted_trial_day": "Day 7 (May 5, 2026)", "uploaded_box_pt": "2026-05-05T16:29:48", "uploader": "Someone", "pages": 1, "size_bytes": 285192, "source_pdf": "PX-15.pdf", "pdf_url": "https://media.mts-in.com/PX-15.pdf", "body_markdown": "# PX 15 — Dec 8, 2015 Altman → Brockman, \"thoughts? from elon\"\n\n> A one-page Sam Altman email to Greg Brockman (\"thoughts? from elon\") forwarding a three-paragraph draft mission statement framing OpenAI as a non-profit AI research company \"unencumbered by an obligation to generate financial returns\" — sent ~3 days before OpenAI's December 11, 2015 public launch.\n\n## Document type\n**Email, plain text, single message.** From Sam Altman <sam@ycombinator.com> to Greg Brockman, sent 12/8/2015 6:14:48 PM, subject \"thoughts? from elon\". Bates OPENAI_MUSK00016993.\n\n## Logistics\n- **Trial admission:** Day 7 (May 5, 2026 — Brockman defense direct + Mr. Molo recross + DX 1252; Wu deposition).\n- **Box upload:** 2026-05-05 16:29:48 PT — late-Day-7 batch (uploaded with PX 207, the 27 MB Day 7 compilation).\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~278 KB, 1 page.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `015.pdf`.\n\n## Transcribed text\n\n> **From:** Sam Altman <sam@ycombinator.com> [on behalf of Sam Altman]\n> **Sent:** 12/8/2015 6:14:48 PM\n> **To:** Greg Brockman\n> **Subject:** thoughts? from elon\n>\n> OpenAI is a non-profit artificial intelligence research company with the goal of advancing digital intelligence in the way that is most likely to benefit humanity as a whole, unencumbered by an obligation to generate financial returns.\n>\n> The underlying philosophy of our company is to disseminate AI technology as broadly as possible as an extension of all individual human wills, ensuring, in the spirit of liberty, that the power of digital intelligence is not overly concentrated and evolves toward the future desired by the sum of humanity.\n>\n> The outcome of this venture is uncertain and the pay is low compared to what others will offer, but we believe the goal and the structure are right. We hope this is what matters most to the best in the field.\n\n## Commentary\n\nPX 15 is a foundational plaintiffs' exhibit on the **founding mission** — the language Altman is forwarding (with the subject line attributing the draft to Musk) tracks closely with OpenAI's December 11, 2015 launch announcement: \"non-profit,\" \"advancing digital intelligence,\" \"benefit humanity as a whole,\" and the explicit \"**unencumbered by an obligation to generate financial returns**\" disclaimer. The email is contemporaneous evidence that, three days before launch, the founders were circulating Musk's own framing of the company as a non-profit. Plaintiffs use this kind of pre-launch text as \"founding documents\" support for the [[Key Themes|\"stole a charity\"]] theory — the same nonprofit framing that Brockman's Aug 21, 2017 journal ([[PX 151]]) and Sept 12, 2017 phone-call notes ([[PX-1250-52|PX 154]]) would later urge be preserved as the \"[[Brockman Journal|moral high ground]].\" See [[Day 7|Day 7 digest]] for the late-Day-7 plaintiffs' upload window into which PX 15 lands.\n\n---\n*See also:* [[Day 7|Day 7 digest]] · [[Greg Brockman]] · [[Sam Altman]] · [[Brockman Journal]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-1503", "exhibit": "PX 1503", "party": "Plaintiffs", "type": "Tweet", "admitted_trial_day": "Day 4 (April 30, 2026)", "uploaded_box_pt": "2026-04-30T14:42:53", "uploader": "Someone", "pages": 1, "size_bytes": 54335, "source_pdf": "PX-1503.pdf", "pdf_url": "https://media.mts-in.com/PX-1503.pdf", "body_markdown": "# PX 1503 — Feb 17, 2020 Musk tweet \"OpenAI should be more open imo\"\n\n> Single-line Elon Musk tweet from February 17, 2020: \"**OpenAI should be more open imo.**\" 125 Reposts, 14 Quotes, 2,051 Likes, 20 Bookmarks at the captured snapshot.\n\n## Document type\n**Tweet (single post screenshot, 1 page).** Posted by @elonmusk, 3:14 PM · Feb 17, 2020. Bates 2024MUSK-0001681. The exhibit sticker on the public copy is green (\"DEFENDANT A\") and labeled \"MDX-1503\"; per Pretrial Order No. 1 admission lists, the canonical exhibit ID for this trial-day batch is **PX 1503**.\n\n## Logistics\n- **Trial admission:** Day 4 (April 30, 2026) — Musk cross + redirect; Birchall direct + cross.\n- **Box upload:** 2026-04-30 14:42:53 PT — Day 4 mid-afternoon batch.\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~54 KB, 1 page.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `1503.pdf`.\n\n## Transcribed text\n\n> **Elon Musk** ✓ 𝕏\n> @elonmusk\n>\n> OpenAI should be more open imo\n>\n> 3:14 PM · Feb 17, 2020\n>\n> 125 Reposts &nbsp;&nbsp; 14 Quotes &nbsp;&nbsp; 2,051 Likes &nbsp;&nbsp; 20 Bookmarks\n\n## Commentary\n\nPX 1503 is a contemporaneous, public Musk statement from **February 17, 2020** — seven months before the more-cited [[Key Themes#\"Captured by Microsoft\" (statute-of-limitations fight)|\"captured by Microsoft\"]] tweet of Sept. 24, 2020 — and is one of the cleanest exhibits on the question of when Musk publicly registered concern about OpenAI's drift from open-source. For the [[Key Themes#The \"OPEN\" in OpenAI|\"OPEN in OpenAI\"]] thread (Musk: \"**The 'OPEN' in OpenAI represents open source**\" on direct, `042826TT.txt:5163`), the seven words on this page do exactly what the cross needs: they fix Musk's own pun-on-the-name framing in 2020. For statute-of-limitations purposes the exhibit cuts both ways — defense will pair it with the September 2020 tweet and Musk's \"I lost confidence\" Nov. 2018 [[DX-844]] line to argue that his complaint was publicly mature by 2020 (\"**you didn't come to court in 2020, did you?**\" `043026TT.txt:2438`); plaintiffs answer (per Musk on direct) that Altman immediately reached out \"to reassure me that OpenAI was on its — was staying on mission as a nonprofit\" (`042926TT.txt:2194`). Used on Day 4 during Musk cross, this is one of the smaller-dollar but higher-leverage exhibits in the SOL fight.\n\n---\n*See also:* [[Day 4|Day 4 digest]] · [[Elon Musk]] · [[Key Themes]] · [[Quotes]]\n"}
{"exhibit_id": "PX-1504", "exhibit": "PX 1504", "party": "Plaintiffs", "type": "Microsoft blog post (printout)", "admitted_trial_day": "Day 4 (April 30, 2026)", "uploaded_box_pt": "2026-04-30T14:42:54", "uploader": "Someone", "pages": 2, "size_bytes": 138524, "source_pdf": "PX-1504.pdf", "pdf_url": "https://media.mts-in.com/PX-1504.pdf", "body_markdown": "# PX 1504 — Sept 22, 2020 Microsoft \"exclusively license GPT-3\" blog post\n\n> A two-page printout of the September 22, 2020 Microsoft blog post by Kevin Scott (CTO) announcing that \"**Microsoft is teaming up with OpenAI to exclusively license GPT-3**\" — plaintiffs' demonstrative of the moment OpenAI's flagship model was first formally walled off behind a Microsoft commercial license.\n\n## Document type\n**Web page printout, two pages.** Source URL: `blogs.microsoft.com/blog/2020/09/22/microsoft-teams-up-with-openai-to-exclusively-license-gpt-3-language-model`. Bates 2024MUSK-0002583–2584; document ID `MDX-1504`.\n\n## Logistics\n- **Trial admission:** Day 4 (April 30, 2026) — used during Musk cross/redirect on the timing of \"phase two\" of Musk's disillusionment narrative (Microsoft commercialization). No prior wiki reference; admission day inferred from Box upload date.\n- **Box upload:** 2026-04-30 14:42:54 PT — Day 4 early-afternoon batch.\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~135 KB, 2 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `1504.pdf`.\n\n## Transcribed text\n\n> **Microsoft teams up with OpenAI to exclusively license GPT-3 language model**\n> blogs.microsoft.com/blog/2020/09/22/microsoft-teams-up-with-openai-to-exclusively-license-gpt-3-language-model\n> September 22, 2020\n>\n> One of the most gratifying parts of my job at Microsoft is being able to witness and influence the intersection of technological progress and impact: harnessing the big trends in computing that have the opportunity to benefit everybody on the planet. Frank's post this morning from Ignite shows just how much progress is happening in many of these areas.\n>\n> Today, the foremost computing trend is undoubtedly artificial intelligence (AI). As we increasingly develop the ability to deploy huge AI models at scale in a way that can be leveraged by all developers and businesses, AI is becoming a platform — an environment upon which folks can build amazing new experiences, just like we've seen happen before with personal computers, mobile devices or the internet.\n>\n> Getting this AI platform off the ground requires unprecedented computing horsepower. So, this May, we expanded upon our ongoing partnership with the world-leading AI research organization OpenAI to announce one of the world's most powerful supercomputers — a custom-designed, Azure-hosted home for training OpenAI's equally massive AI models.\n>\n> Since then, you've probably already seen OpenAI's announcement of their groundbreaking GPT-3 model — an autoregressive language model that outputs remarkably human-like text. GPT-3 is the largest and most advanced language model in the world, clocking in at 175 billion parameters, and is trained on Azure's AI supercomputer.\n>\n> Today, I'm very excited to announce that **Microsoft is teaming up with OpenAI to exclusively license GPT-3**, allowing us to leverage its technical innovations to develop and deliver advanced AI solutions for our customers, as well as create new solutions that harness the amazing power of advanced natural language generation.\n>\n> We see this as an incredible opportunity to expand our Azure-powered AI platform in a way that democratizes AI technology, enables new products, services and experiences, and increases the positive impact of AI at Scale. Our mission at Microsoft is to empower every person and every organization on the planet to achieve more, so we want to make sure that this AI platform is available to everyone — researchers, entrepreneurs, hobbyists, businesses — to empower their ambitions to create something new and interesting.\n>\n> The scope of commercial and creative potential that can be unlocked through the GPT-3 model is profound, with genuinely novel capabilities — most of which we haven't even imagined yet. Directly aiding human creativity and ingenuity in areas like writing and composition, describing and summarizing large blocks of long-form data (including code), converting natural language to another language — the possibilities are limited only by the ideas and scenarios that we bring to the table.\n>\n> Realizing these benefits at true scale — responsibly, affordably and equitably — is going to require more human input and effort than any one large technology company can bring to bear.\n>\n> On that journey, today is only the beginning of the beginning. OpenAI will continue to offer GPT-3 and other powerful models via its own Azure-hosted API, launched in June. While we'll be hard at work utilizing the capabilities of GPT-3 in our own products, services and experiences to benefit our customers, we'll also continue to work with OpenAI to keep looking forward: leveraging and democratizing the power of their cutting-edge AI research as they continue on their mission to build safe artificial general intelligence. That future will be what we make of it — and I believe that we're on the right track.\n>\n> Tags: AI, Ignite 2020\n\n## Commentary\n\nPX 1504 is plaintiffs' \"phase two\" anchor — the public moment in September 2020 when GPT-3, OpenAI's signature model, was contractually walled off behind a Microsoft \"exclusive license.\" The blog headline is itself the proof point (\"**exclusively license GPT-3**\") and plaintiffs use it to argue that the flagship output of the nonprofit's research mission was being routed into a Microsoft commercial product line, contrary to the OpenAI Charter ([[PX 24]]) commitment that the organization's \"primary fiduciary duty is to humanity\" and that its work would be \"broadly distributed.\" Defense reads the same post as an OpenAI-friendly arrangement: the API \"launched in June\" remains available, the language tracks OpenAI's mission, and Microsoft's frame is \"democratizes AI technology, enables new products, services and experiences.\" On cross, Birchall was walked through the September 2020 announcement to establish that the Musk-side Foundation/family office knew and did not object — relevant to the defense statute-of-limitations theory that any breach claim accrued no later than 2020. Cross-reference to the [[Watershed MOU]] and Robert Wu's Day 7 30(b)(6) testimony on the underlying Microsoft economics.\n\n---\n*See also:* [[Day 4|Day 4 digest]] · [[PX 24]] · [[Watershed MOU]] · [[Statute of Limitations]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-151", "exhibit": "PX 151", "party": "Plaintiffs", "type": "Personal journal entry", "admitted_trial_day": "Day 6 (May 4, 2026)", "uploaded_box_pt": "2026-05-04T14:46:05", "uploader": "Someone", "pages": 2, "size_bytes": 187597, "source_pdf": "PX-151.pdf", "pdf_url": "https://media.mts-in.com/PX-151.pdf", "body_markdown": "# PX 151 — Aug 21, 2017 Brockman journal entry (\"get out from Elon\" / \"$1B\")\n\n> Greg Brockman's two-page contemporaneous laptop-journal entry from August 21, 2017, written on the eve of the OpenAI for-profit-vs.-nonprofit decision: \"**This is the only chance we have to get out from Elon**\" / \"**Financially what will take me to $1B?**\" / \"Accepting Elon's terms nukes two things: our ability to choose … and the economics.\"\n\n## Document type\n**Personal journal entry, plain ASCII text, two pages.** Filename embedded in document: `2017-08-21.txt`. Stream-of-consciousness diary kept on Brockman's laptop since 2010 (see [[Brockman Journal]]). Bates OPENAI_MUSK00039292–39293; stamped \"Highly Confidential.\" Three small black-box redactions for \"Personal\" content.\n\n## Logistics\n- **Trial admission:** Day 6 (May 4, 2026). Used by **Mr. Kry** on cross-examination of Greg Brockman. Re-read during Day 7 Mr. Kry redirect.\n- **Box upload:** 2026-05-04 14:46:05 PT — Day 6 mid-afternoon batch (clustered with PX 161, PX 241, PX 390).\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~183 KB, 2 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `151.pdf`.\n\n## Transcribed text\n\n> **File Name: 2017-08-21.txt**\n>\n> Ok so what do I *really* want? I want to want to be an engineer. But I think now is a crazy shot to be the one in charge and to step up to the challenge. We might succeed, truly. It is already a different game from anyone I know.\n>\n> Such upside. **This is the only chance we have to get out from Elon.** Is he the \"glorious leader\" that I would pick? We truly have a chance to make this happen. **Financially what will take me to $1B? Wo**\n>\n> I'm definitely afraid of the switching cost.\n>\n> On the other hand, [Redacted - Personal], this is probably the most interesting thing you'll do. So given that, when will I ever have a chance to be free and truly own my destiny?\n>\n> We lose out on the Jim Keller route. If we need to build our own hardware it's not so great. Though honestly if we need to do it then we need to do it. I firmly believe in our ability to hire there and not in Elon's ability to build the right hardware without Ilya.\n>\n> **Accepting Elon's terms nukes two things: our ability to choose (though maybe we could overrule him) and the economics.** Some chance that rejecting Elon will actually lose us Sam. We'll find out tomorrow if doing an override all the way through is palettable.\n>\n> I'm taking on a lot of pain and suffering. To get 10% of the thing at a billion, we're talking $100m. [Redacted - Personal] [Redacted - Personal] and so yeah why am I going to accept that pain?\n>\n> i've been in charge for 1.5-2 years now. i built this team from nothing. elon helped get it started. but we've really figured out what we're trying to do now.\n\n## Commentary\n\nPX 151 is plaintiffs' single most-quoted journal passage and the document Mr. Kry built his Day 6 cross around. The lines **\"This is the only chance we have to get out from Elon\"** and **\"Financially what will take me to $1B?\"** were repeatedly put to Brockman as contemporaneous proof that the August–September 2017 negotiation was, on the founders' side, primarily about removing Musk and capturing economic upside, not — as Brockman testified — about preventing unilateral AGI control. Plaintiffs' read (per Molo's recross): \"It's just an after-the-fact manufactured excuse for the very honest emotion that you expressed in the journal at the time.\" Brockman's defense read on the stand: \"It's an expression of frustration, not a plan\" — and the \"$1B\" line was a private mental ceiling above which the with-Musk-vs.-without-Musk choice would be financially indifferent. The \"Accepting Elon's terms nukes … the economics\" line is what Mr. Kry returned to repeatedly when contrasting the journal with [[PX-157|PX 157's]] \"happy to give up on the equity\" representation to Musk. See [[Brockman Journal]] for the full sequence of journal entries (Aug 18 PX 154, this entry, Sept 12 PX 154, Nov 5 [[DX-1252]], Nov 6 [[PX-161]], Nov 12 PX 163) and [[Day 6|the Day 6 digest]] for the contemporaneous testimony at ~11:23 PT.\n\n---\n*See also:* [[Brockman Journal]] · [[Day 6|Day 6 digest]] · [[Day 7|Day 7 digest]] · [[PX-161]] · [[PX-157]] · [[Greg Brockman]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-152", "exhibit": "PX 152", "party": "Plaintiffs", "type": "Email thread", "admitted_trial_day": "Day 2 (April 28, 2026)", "uploaded_box_pt": "2026-04-28T14:52:00", "uploader": "Someone", "pages": 2, "size_bytes": 272801, "source_pdf": "PX-152.pdf", "pdf_url": "https://media.mts-in.com/PX-152.pdf", "body_markdown": "# PX 152 — Aug 28, 2017 Musk reply to Zilis (\"I've had enough\")\n\n> Shivon Zilis's seven-question debrief to Musk after a weekend conversation with Brockman and Sutskever — and Musk's three-week-early \"I've had enough\" response that prefigures the [[PX 157]] \"final straw\" email by twenty-three days.\n\n## Document type\n**Email thread, plain text, two messages.** Zilis → Musk (cc Sam Teller), Aug 28, 2017 12:01 AM, subject \"OpenAI notes.\" Musk → Zilis (cc Sam Teller) reply at 28 Aug 2017 00:08:50 -0700, same subject. Bates 2024MUSK-0005521–0005522.\n\n## Logistics\n- **Trial admission:** Day 2 (April 28, 2026). The wiki reference index identifies this exhibit specifically as \"PX 152 — Aug 28, 2017 Musk reply (had enough).\"\n- **Box upload:** 2026-04-28 14:52:00 PT — Day 2 mid-afternoon batch (one second after [[PX 14]]).\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~266 KB, 2 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `152.pdf`.\n\n## Transcribed text\n\n> **From:** Elon Musk <erm@spacex.com>\n> **To:** Shivon Zilis <shivon@openai.com>\n> **Cc:** Sam Teller <steller@spacex.com>\n> **Subject:** Re: OpenAI notes\n> **Date:** Mon, 28 Aug 2017 00:08:50 -0700\n> **Importance:** Normal\n>\n> This is very annoying. Please encourage them to go start a company. I've had enough.\n\n> **From:** Shivon Zilis <shivon@openai.com>\n> **To:** Elon Musk <erm@spacex.com>\n> **Cc:** Sam Teller <steller@spacex.com>\n> **Subject:** OpenAI notes\n> **Date:** Mon, Aug 28, 2017 at 12:01 AM\n>\n> Elon,\n>\n> As I'd mentioned, Greg had asked to talk through a few things this weekend. Ilya ended up joining, and they pretty much just shared all of what they are still trying to think through. This is the distillation of that random walk of a conversation… came down to 7 unanswered questions with their commentary below. Please note that I'm not advocating for any of this, just structuring and sharing the information I heard.\n>\n> **1. Short-term control structure?**\n> -Is the requirement for absolute control? They wonder if there is a scenario where there could be some sort of creative overrule provision if literally everyone else disagreed on direction (not just the three of them, but perhaps a broader board)?\n>\n> **2. Duration of control and transition?**\n> -*The* non-negotiable seems to be an ironclad agreement to not have any one person have absolute control of AGI if it's created. Satisfying this means a situation where, regardless of what happens to the three of them, it's guaranteed that power over the company is distributed after the 2-3 year initial period.\n>\n> **3. Time spent?**\n> -How much time does Elon want to spend on this, and how much time can he actually afford to spend on this? In what timeframe? Is this an hour a week, ten hours a week, something in between?\n>\n> **4. What to do with time spent?**\n> -They don't really know how he prefers to spend time at his other companies and how he'd want to spend his time on this. Greg and Ilya are confident they could build out SW / ML side of things pretty well. They are not confident on the hardware front. They seemed hopeful Elon could spend some time on that since that's where they are weak, but did want his help in all domains he was interested in.\n>\n> **5. Ratio of time spent to amount of control?**\n> -They are cool with less time / less control, more time / more control, but not less time / more control. Their fear is that there won't be enough time to discuss relevant contextual information to make correct decisions if too little time is spent.\n>\n> **6. Equity split?**\n> -Greg still instinctually anchored on equal split. I personally disagree with him on that instinct and he asked for and was receptive to hearing other things he could use to recalibrate his mental model.\n> -Greg noted that Ilya in some ways has contributed millions by leaving his earning potential on the table at Google.\n> -One concern they had was the proposed employee pool was too small.\n>\n> **7. Capitalization strategy?**\n> -Their instinct is to raise much more than $100M out of the gate. They are of the opinion that the datacenter they need alone would cost that so they feel more comfortable raising more.\n>\n> **Takeaways:**\n> Unsure if any of this is amenable but just from listening to all of the data points they threw out, the following would satisfy their current sticky points:\n> -Spending 5-10 hours a week with near full control, or spend less time and have less control.\n> -Having a creative short-term override just for extreme scenarios that was not just Greg / Sam / Ilya.\n> -An ironclad 2-3yr minority control agreement, regardless of the fates of Greg / Sam / Ilya.\n> -$200M-$1B initial raise.\n> -Greg and Ilya's stakes end up higher than 1/10 of Elon's but not significantly (this remains the most ambiguous).\n> -Increasing employee pool.\n\n## Commentary\n\nPX 152 is plaintiffs' single cleanest demonstration that **Musk's \"final straw\" was not sudden** — three weeks before [[PX 157]] he was already telling his information conduit at OpenAI that the founders should \"go start a company\" and \"I've had enough.\" Defense's read is the mirror image: read the Zilis memo, not the Musk reply, and it is **Musk** who is being asked for fewer hours and less control while the founders offered an \"ironclad 2-3 year minority control agreement\" — the same ironclad-against-anyone's-absolute-control red line that Zilis testified to on Day 8 (see [[Key Themes]] §\"ironclad agreement\"). The \"Greg and Ilya's stakes end up higher than 1/10 of Elon's\" line is a useful number for plaintiffs: the founders' aspirational equity at this date was small, and the eventual ~$30B Brockman stake (see [[Key Themes]]) is downstream of Musk's refusal here, not of an earlier deal. The contemporaneous Brockman entries from this same week — [[PX 151]] (Aug 21) and the [[Brockman Journal]] generally — should be read alongside PX 152 to triangulate what the founders actually wanted on the eve of Musk's withdrawal.\n\n---\n*See also:* [[Day 2|Day 2 digest]] · [[PX 151]] · [[PX 157]] · [[Brockman Journal]] · [[Shivon Zilis]] · [[Greg Brockman]] · [[Ilya Sutskever]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-153", "exhibit": "PX 153", "party": "Plaintiffs", "type": "Email", "admitted_trial_day": "Day 2 (April 28, 2026)", "uploaded_box_pt": "2026-04-28T14:51:59", "uploader": "Someone", "pages": 3, "size_bytes": 259883, "source_pdf": "PX-153.pdf", "pdf_url": "https://media.mts-in.com/PX-153.pdf", "body_markdown": "# PX 153 — Sept 11, 2017 Birchall → Musk, \"OpenAI Cap Table\" (Brockman/Sutskever proposal)\n\n> Jared Birchall's September 11, 2017 cover email to Elon Musk forwarding a \"more user friendly version of the cap table that Ilya and Greg are proposing\" — the proposed founder cap table for the for-profit OpenAI conversion under negotiation: **Elon 51.20%, Sam/Ilya/Greg ~11% each (3.33% paid + 7.68% granted), employee pool 7.47%**.\n\n## Document type\n**Email + spreadsheet attachment.** Page 1 is Birchall's cover note from `jared@excession.com` to Musk on 9/11/2017, attaching `OpenAI Cap Table.xlsx`. Page 2 is the standard \"File Produced Natively\" placeholder for the spreadsheet. Page 3 is the printed contents of the spreadsheet — a two-table cap-table with shareholder allocations and an employee-allocation breakdown. Bates EXMF-0003257–3258, with the spreadsheet dump on the third page.\n\n## Logistics\n- **Trial admission:** Day 2 (April 28, 2026 — openings + Musk direct begins). Per the chunk metadata: \"PX 153 — referenced in Day 2.\" Also re-used Day 3 (042926TT.txt:1255–1336) and Day 6 (5/4/2026 Testimony @ ~10:25 PT) per [[Brockman Journal|Brockman Journal context]] and [[Key Themes]].\n- **Box upload:** 2026-04-28 14:51:59 PT — Day 2 mid-afternoon batch.\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~254 KB, 3 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `153.pdf`.\n\n## Transcribed text\n\n> **From:** Jared Birchall <jared@excession.com>\n> **To:** Elon Musk <erm@spacex.com>\n> **Subject:** OpenAI Cap Table\n> **Date:** 9/11/2017 11:25:40 PM\n> **Attachments:** OpenAI Cap Table.xlsx\n>\n> I've attached a more user friendly version of the cap table that Ilya and Greg are proposing.\n\n[Page 2: \"File Produced Natively\"]\n\n> **OpenAI Cap Table — printed contents:**\n>\n> | Shareholder | Initial Shares | % |\n> |---|---:|---:|\n> | **Elon Investment** | 100,000,000 | 51.20% |\n> | **Sam Investment** | 6,500,000 | 3.33% |\n> | Sam grant | 15,000,000 | 7.68% |\n> | **Ilya Investment** | 6,500,000 | 3.33% |\n> | Ilya grant | 15,000,000 | 7.68% |\n> | **Greg Investment** | 6,500,000 | 3.33% |\n> | Greg grant | 15,000,000 | 7.68% |\n> | John grant | 3,906,250 | 2.00% |\n> | Woj grant | 3,906,250 | 2.00% |\n> | **Employees** | 14,589,843 | 7.47% |\n> | **Pool** | 8,410,156 | 4.31% |\n> | **Total** | **195,312,499** | **100.00%** |\n>\n> **Initial thoughts for Employees** [employee-by-employee allocation table; representative entries]:\n>\n> rafal 1,171,875 (0.6%); jakub 1,171,875 (0.6%); szymon 1,171,875 (0.6%); marcin 1,171,875 (0.6%); jonas, vicki, csh, bob, pw, cberner, scott, jie, tim, josh, alec, prafulla, durk @ 585,937 (0.3% each); igor, harri, jack, paul, dario, geoffrey @ 195,312 (0.1% each); bradly, richardchen, yura, rein, brooke, oleg, vicki pfau, karthik, maciek, alex nichol, chris @ 97,656 (0.05% each); gail, kate @ 19,531 (0.01% each). Total: 14,589,844 / 7.47%.\n\n## Commentary\n\nPX 153 is the **September 2017 cap-table proposal** — the document that crystallizes the [[Key Themes|\"unequivocal control\"]] negotiation. On the page, in Brockman/Sutskever's proposal as forwarded by Birchall to Musk, **Musk receives 51.20%** of the for-profit OpenAI's equity in exchange for a $100M cash investment, the founders each receive ~11% (3.33% paid + 7.68% sweat-equity grant), and a 7.47% employee pool is divvied up by name and dollar (down to \"gail\" and \"kate\" at 0.01%). This is the cap table Brockman would, three days later, agree was on the table when he was writing the [[PX 151|\"only chance to get out from Elon … take me to $1,000,000,000\"]] entry in his journal; it is also defense's strongest \"Musk wanted majority control\" exhibit (Cohen/Savitt cross of Musk used it as \"you would have **51 percent of the company**\" — see [[Key Themes]] § \"Unequivocal control\"). Plaintiffs use it to anchor the **[[Key Themes|\"Greg Brockman's contributions ('zero')\"]]** theme: Brockman's 11.01% in the proposed cap table consists of a $6.5M check (which he never wrote) plus 7.68% in granted equity — the structural antecedent of the **~$30 billion** stake he ultimately did receive in 2018 having paid $0 for it. Re-used on Day 3 (042926TT.txt:1255–1336) on Musk cross and Day 6 (~10:25 PT) on Brockman cross.\n\n---\n*See also:* [[Day 2|Day 2 digest]] · [[Day 3|Day 3 digest]] · [[Day 6|Day 6 digest]] · [[PX 151]] · [[PX 157]] · [[Greg Brockman]] · [[Jared Birchall]] · [[Brockman Journal]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-154", "exhibit": "PX 154", "party": "Plaintiffs", "type": "Personal journal entry", "admitted_trial_day": "Day 6 (May 4, 2026)", "uploaded_box_pt": "2026-05-04T14:46:10", "uploader": "Someone", "pages": 9, "size_bytes": 566463, "source_pdf": "PX-154.pdf", "pdf_url": "https://media.mts-in.com/PX-154.pdf", "body_markdown": "# PX 154 — Brockman journal, September 12, 2017 (Sam call / Elon call / Ilya summary email)\n\n> Greg Brockman's contemporaneous laptop journal entry from September 12, 2017 — capturing back-to-back calls with Sam Altman and Elon Musk during the August–September 2017 control negotiation, plus Ilya Sutskever's same-night recap email to Musk that became the textual basis for the [[PX-157|\"Honest Thoughts\"]] letter eight days later.\n\n## Document type\n**Personal journal entry.** File header: `File Name : 2017-09-12.txt`. Stream-of-consciousness diary kept on Brockman's laptop since 2010, internally Bates-numbered (2024MUSK series; OPENAI_MUSK00039364–00039372). Mixes Brockman's running self-questioning with near-verbatim dialogue transcripts of phone calls and a long quoted email thread with Musk. Marked \"HIGHLY CONFIDENTIAL.\"\n\n## Logistics\n- **Trial admission:** Day 6 (May 4, 2026) — Russell direct + cross; Brockman cross by Mr. Kry.\n- **Box upload:** 2026-05-04 14:46:10 PT.\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~553 KB, 9 pages. Bates 1250.x range (Brockman journal).\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `154.pdf`.\n\n## Transcribed text\n\n**[Page 1: file label]** `File Name : 2017-09-12.txt`\n\n**[Page 2: Brockman's \"latest thought process\"]**\n\n> latest thought process:\n>\n> - i'm not happy with the way he's steamrolling sam, and that he demands to be ceo so that it can be clear that he has all the control. he just wants it and it'll make sam's job harder.\n> - i'm starting to feel more of the fighting spirit, and like the idea of being able to start our own thing. it's no longer a fight about money, it's a fight over what kind of company we want to build. he's going to be the ceo and run engineering+research but knows nothing about these right now. maybe he'll become a formidable force. but right now he's definitely not there.\n> - what do i want really? i would like to succeed at building the agi. i succeeded at dota 1v1, and that's very clearly mine. i would like to go higher, to be respected for what i do. he's stepping on our throats a bit here and the beatings continue until morale improves.\n> - honestly i just defer to him and say nice things and never push back on anything. i wouldn't respect me very much in his shoes.\n> - so yeah, i think walking away despite the CEO point isn't crazy. **do we really want to be the people who bring elon to control of the AGI.** the flip side is he is fucking famous. he's got the resources. and he is smart and has a really good sense of what to work on. is it true and has he been that important for us?\n\n> ** sam call\n>\n> - sam and e have a call. e says he must be founding ceo but happy to transition the title when greg/ilya want.\n>\n> 1 minute ago / i did not think we were gonna get to where we are now / yesterday i was sure the band was lost / now / heh now / so now we know who's the real drama queen ;) / now / funny how our mentalities shift / going to do a loan to ilya for maybe up to 8M / anyway, do i wanna keep the band together? answer: yes, i do.\n\n**[Page 3: Brockman/Sutskever's \"asks for elon\" + the Elon call begins]**\n\n> asks for elon:\n> - let ilya get to the 20M with sam and me. would be nice to keep the extra 2M grant but understood if not. this\n>\n> ** elon call\n>\n> our asks:\n> - close out equity: first, want to check that $10M investment is your comfort limit. 20M for greg/sam, 20M for ilya, with loan + founders grant. early liquidity?\n> - as CEO, you'll be very visible. if we do good and important work, what are your thoughts on credit for us?\n> - what does your research and eng role look like?\n>\n> - elon: talked to sam a fair bit. did you have a chance to connect?\n> - gdb: yep.\n> - elon: he was pretty hung up on ceo thing less night, seemed last so now heading into din. need at beginning who is running the company, sam was saying everyone would know it's elon, e said don't think should be diff from reality and title. **i don't like being ceo of anything. intent to not be ceo of tesla once done with model 3** in particular, he was hungup, (he said showstopper).\n> - gdb: thought you talked about swapping the title.\n> - elon: not a certainty, not like he is guaranteed the job. if the consensus view of the senior team is that they want him as the ceo, inclusive as elon, right guy to be ceo, would support. don't want to be ceo for the sake of it, fucking hate being ceo, first startup tried extremely hard, did not want to be ceo, brother was ceo, wanted to be basically programming, vcs invested and hired a ceo, discovered that if you are not the ceo, people can put some bozo in charge who make you do crazy things that suck. so then after that was like goddamn it, i don't want to be ceo, if i'm stuck on the ship and can't get off the ship, then gotta be ceo. given the importance of this particular issue, can't see why it'd make sense -- it'd be highest priority thing. arguably is right now but have short-term obligations i can't release. this is not a case of sam gets to be ceo automatically. down the road.\n> - gdb: what would that look like?\n> - elon: this is wearing on me. if sam wants to join that's great, join as president/coo, great. **if you wanna do something with me, i've had enough of dancing around the bush. not doing succession**\n\n**[Page 4: continued Elon call — equity terms]**\n\n> planning. i'm gonna go do my own my thing. am i clear?\n> - gdb: very clear.\n> - elon: have enough shit to deal with without this on my plate. dealing with someone's insecurities not something i want to do. can be this path or another path. not going to engage in complicated.\n> - gdb: understood. let's close out our conversation.\n> - ilya: so greg willing to do loan for 5.5M, securitized by YC stock. 2.5 from me. 12M.\n> - elon: what's the 2.5M that's not YC?\n> - ilya: i have 7M, half in tesla stock.\n> - elon: haha really? thank you. vote of confidence.\n> - ilya: can't sell it. i talk to you a lot. it's done very well for me.\n> - elon: i support it, when all the cards are on the table, can explain it to the rest of the team.\n> - ilya: gave up google cash, got a lot of YC stock.\n> - ilya: if i understood correct: 12M grant, 2.5M cash, 5.5M willing to do securitized YC stock (10x larger than the ones from others).\n> - elon: sounds good. would ask jared what occurred with neuralink. for any pledging exactly what assets are pledged. that works for me.\n> - ilya: to make sure on same page: current picture 10M, 10M, 12M, $2.5M from me, and $5.5M from greg securitized against YC stock.\n> - elon: think it's something that can be explained.\n> - ilya: question: i'm taking out the loans. would prefer to return sooner-ish if it's possible. what are your views on early liquidity?\n> - elon: yes, absolutely. the way it works at spacex. have been able to get the best of all possible worlds. the liquidity of a public company without all the pain. e.g. stock moving randomly up or down. can be a mood thermometer. at spacex external valuation done every 6 months, liquidity event 2x a year. spacex participates in stock buybacks every financing round. want to keep # of shareholders under control. will do the same thing here. aspire at least once a year (maybe 2x) can sell stock at a valuation determined by an outside company.\n> - gdb: at spacex, was a while into the company right?\n> - elon: first 5 years, trying not to die. the first liquidity event where people able to sell 2008. 2006 brink of destruction. to some measure of prosperity. i suspect 1-2 years. could accelerate that if\n\n**[Page 5: continued — security, AI, and Putin/Xi]**\n\n> there.\n> - gdb: assume $10M is comfort limit.\n> - elon: subsequent rounds, i'll invest there.\n> - gdb: you'll be in spotlight,\n> - elon: i would like to be in the spotlight the least amount possible. would like to highlight you. anything you can think of.\n> - gdb: means quite a bit.\n> - elon: not looking for awards/praise/anything else. **my reasons for doing this are because my #1 thing keeping me awake at night is AGI. what the fuck are we gonna do. sure hope this ends up mitigating that.** #1 buzzkill in life. don't want any awards, any anything. nothing. saudi's offered $1M to fly there and win some env award. if there's anything you can think of that dimishes me and highlights\n> - gdb: what would your role look like, as you scale up?\n> - elon: want to amp up security.\n> - gdb: [Redacted - Personal]\n> - elon: **playtime is over on this matter. when putin and president xi describe this.**\n> - gdb: know how to do infosec. less so the physical security.\n> - elon: amping that up for myself. value of that, $9B of short sale, for one of those guys alone to get the info, may be worse than russian secret service. countersurveillance gonna dial up massively across the board. i know at least some of the right people to talk to in gov, spacex did a fair bit of classified work. gonna need to treat like advanced weapons tech.\n> - gdb: (reminds quesiton)\n> - elon: needs to gain a much better understanding on AI on a fundamental level. want to get to the same level of understanding as of rockets + electric cars. in order to make right decisions quite fundamental. a lot of info to upload. way it worked at spacex/tesla from beginning involved in countless design meetings of every aspect of the rocket + car. not something that can be created overnight. present, listen and ask questions, read whatever i can, look at the software, understand the math, understand the fundamental algo.\n> - ilya: are you considering getting hands on exp with writing the software?\n> - elon: think i should. wrote all the code of zip2, with lousy\n\n**[Page 6: continued — talent war, AGI fear, and the \"moral high ground\" line]**\n\n> computing tech. lots of video games. stopped programming seriously in mid- to late-20s. gotta do some of that.\n> - ilya: think it'd be really helpful.\n> - elon: gotta close loop around ground truth. gonna make my brain hurt.\n> - gdb: definitely made my brain hurt, but not that hard if you really dig in.\n> - elon: gotta do it. kind of thing i like. like technology. load up the old brain. huge part is gonna be a talent war. how do we convince the best people to join us.\n> - gdb: if you and sam get more involved, will be unstoppable.\n> - ilya: new structure will help a lot too.\n> - elon: alright sounds good. game is afoot. gonna be battle.\n> - gdb: pretty soon will remember the days when enemy was just DM.\n> - elon: hah. don't want to pave road to hell with good intentions.\n> - ilya: don't create the AGI before making it act in our best interests. should be fundamental tenant. especially once we have the 10GW datacenter, can do a lot of smart things, things they do should be good for us.\n> - elon: i feel fear. don't know about you guys.\n> - gdb: used to feel born 10 years too late. missed the 90s. just grateful to have an opp to help this go well.\n> - elon: hard to believe this is real. reality is increasingly surreal.\n> - ilya: let's hope it doesn't take a negative turn.\n> - elon: don't approach with too heavy a heart. it's a serious matter. if things do go awry, enjoy these years on earth.\n> - gdb: mix of hedonistic parties and heads down writing code.\n> - elon: have a few parties, celebrate the wins. reasonably good time along the way.\n> - gdb: over upcoming weeks, how much of your time should we plan for?\n> - elon: **coming weeks, top priority. gotta figure out how do we transition from non-profit to something which is essentially philanthropic endeavor and is B-corp or C-corp or something. must tell the story and not lose moral high ground. absolutely vital.**\n> - gdb: yep.\n> - elon: need to understand B-corp situation.\n> - ilya: i have *some* thoughts. one idea keep non-profit and have C-corp. but i find this approach less appealing because in some sense C-corp not beholden. one formulation: mission is to minimize\n\n**[Page 7: continued — naming, structure, and the Ilya recap]**\n\n> ex risk by building friendly AGI. could be a good way to go. like it more as new entity beholden to the mission.\n> - elon: agree, does sound better. never even heard of a B-corp until sam brought it up. it does sound like right move. would not shut down the non-profit, should still exist in some form.\n> - ilya: no opinion, as long as the main entity has something fundamentally philanthropic.\n> - elon: any prefs on name?\n> - ilya: openai. continuity of the mission. it's all the same.\n> - elon: i agree, think that makes sense. well, i'm pretty excited about doing this with you.\n> - ilya: all that's left is the minor thing of actually doing it.\n> - elon: alright cool. i'm gonna figure out the details of the whole B-corp thing tomorrow and get that process under way. and then let's just stay in frequent touch. make this happen as quickly as possible.\n> - gdb: super excited.\n> - ilya: most drastic thing that can possibly be done.\n> - elon: call quality is degrading. maybe too many people spying on the call.\n>\n> ** notes from ilya\n>\n> Had another call with Elon yesterday.\n>\n> I no longer remember the precise order of the topics in the call. Roughly speaking:\n>\n> We started by Greg asking about the Sam situation. Asked about the status. Elon said that he's OK with Sam being the CEO if the entire senior team, inclusive of Elon, would support Sam as CEO. However, it is very important to Elon to start as the CEO of this thing, because **he wants to send a message to the team that he's truly in charge here.** He said it very explicitly. He said that at some point in the future, he'd be OK with Sam being CEO, if he felt that Sam is the right person for the job. Emphasized that Sam is not guaranteed the job. Greg asked, how do you imagine that would look like? **And Elon exploded. Said, \"I'm at the end of my rope here, I don't want to play games anymore. If you're not happy, go start your company with Sam, and I'll do it on my own. Am I being clear\". Greg said, \"extremely clear\". I found this of the conversation to be so distasteful that I had a knot in my stomach writing these notes. Very sad.**\n>\n> At the same time, he strongly emphasized how he hates being CEO, how he wants to be an engineer, how he doesn't like it, etc, etc.\n>\n> On credit, he said that he doesn't want credit, that he's happy to do anything at all that we'd ask him in order to help highlight Greg and I. He said that he'll do anything to reduce his own credit and increase ours.\n>\n> We covered equity. I asked if he'd be OK with me getting a 5.5M loan from Greg securitized by my YC stock. He said, all I care is that there's a clear story to the team. I said, when I joined OpenAI, I negotiated a lange grant of YC stock because of the Google money I stood to lose. It is 10x larger than the second larger grant, and is more vested because half of it was vesting conditioned on me joining, as a signing bonus. He asked me about the other money I have, I said it's 7m, half of which in\n\n**[Page 8: continued — the long Musk reply on equity & control]**\n\n> Tesla stock. That was a small nice surprise.\n>\n> I asked about whether I could sell my shares relatively early, in order to pay back the loans I was going to take. He said basically sure.\n>\n> We then had a bunch of pleasant conversation about the structure (C-corp vs B-corp, non-profit vs not), talked about the need of safety, the name (keep \"OpenAI\"). Etc. All went well.\n>\n> I then sent Elon an email summary of the notes from the conversation, to consolidate our discussions in writing re equity and control.\n>\n> **He wrote us a long reply, where he took back the previous ideas on control that were on the table. Specifically, he said that instead of him starting with 3 board seats and Greg/Sam/I have a board seat each, he said he'd start with four board seats. Not three. Because he wants to start with unequivocal control in the new org.**\n>\n> For the longest time, I used to take anything Elon said at face value. I sincerely believed that he doesn't care about credit, that he doesn't want to be CEO, etc, etc. **But I am having less confidence in this right now, which worries me, because I now have less confidence that he'll voluntarily diminish his control to the 25% that he's been promising -- because he can just change his mind and there's nothing we would be able to do about it.**\n>\n> ** sam call\n> talk to sam, he says elon's tone is very different from on their call. expected him to back off. will talk tomorrow and hopefully get there.\n>\n> ** email exchange with elon\n>\n> Sounds good. The three common stock seats (you, Greg and Sam) should be elected by common shareholders. They will de facto be yours, but not in the unlikely event that you lose the faith of a huge percentage of common stockholders over time or step away from the company by choice.\n>\n> I think that the Preferred A investment round (supermajority me) should have the right to appoint four (not three) seats. **I would not expect to appoint them immediately, but, like I said I would unequivocally have initial control of the company,** but this will change quickly. The rough target would be to get to a 12 person board (probably more like 16 if this board really ends up deciding the fate of the world) where each board member has a deep understanding of technology, at least a basic understanding of AI and strong & sensible morals.\n>\n> Apart from the Series A four and the Common three, there would likely be a board member with each new lead investor/ally. However the specific individual new board members can only be added if all but one existing board members agrees. Same for removing board members.\n>\n> There will also be independent board members we want to add who aren't associated with an investor. Same rules apply: requires all but one of existing directors to add or remove.\n>\n> **I'm super tired and don't want to overcomplicate things, but this seems approx right. At the sixteen person board level, we would have 7/16 votes and I'd have a 25% influence, which is my min comfort level. That sounds about right to me. If everyone else we asked to join our board is truly against us, we should probably lose.**\n>\n> As mentioned, my experience with boards (assuming they consist of good, smart people) is that they are rational and reasonable. There is basically never a real hardcore battle where an individual board vote is pivotal, so this is almost certainly (sure hope so) going to be a moot point.\n\n**[Page 9: closing — Sutskever's recap email back to Musk]**\n\n> As a closing note, I've been really impressed with the quality of discussion with you guys on the equity and board stuff. I have a really good feeling about this.\n>\n> Lmk if above seems reasonable.\n>\n> Elon\n>\n> > On Sep 12, 2017, at 11:44 PM, Ilya Sutskever <[redacted]> wrote:\n> >\n> > Hi Elon,\n> > To summarize our understanding of the current state:\n> >\n> > On equity:\n> > Greg: 10M grant/10M investment\n> > Sam: 10M grant/10M investment\n> > Ilya: 12M grant/2.5M investment + 5.5M loan from Greg securitized by Ilya's YC vested stock, from his work at OpenAI\n> >\n> > On control:\n> > 3 board seats for Elon, 1 board seat each for Ilya/Greg/Sam. Plan to expand board over time through unanimous consent of current board, up to 12 people. Details TBD on tiebreak and whether the directors are statically allocated.\n> >\n> > On credit:\n> > Elon is extremely happy to help Greg and Ilya get credit for their work -- and is very open to any ideas Greg and Ilya may have.\n> >\n> > Looking forward to the conclusions of your conversation with Sam tomorrow.\n> >\n> > Ilya\n\n## Commentary\n\nPX 154 is the **textual seed** of the [[PX-157|\"Honest Thoughts\" letter]] eight days later — every concern Sutskever and Brockman raise on Sept 20 (\"you've shown to us that absolute control is extremely important to you,\" \"you needed to be CEO so that everyone will know that you are the one in charge,\" \"you wanted to retain absolute control as the company makes genuine progress towards AGI\") is rehearsed here, in real time, on Sept 12. **Plaintiffs' read** centers on Brockman's \"**do we really want to be the people who bring elon to control of the AGI**\" line and Musk's own recorded statement \"**must tell the story and not lose moral high ground. absolutely vital.**\" — the exhibit Mr. Molo put to Brockman on cross at ~09:22 PT on Day 7 to argue that even Musk understood the for-profit conversion required keeping the public-facing nonprofit narrative intact. **Defense's read** uses the same passage (and Sutskever's recap email at the end) for the [[Key Themes|\"unequivocal control\"]] thread: Musk's own words (\"**i would unequivocally have initial control of the company**\"; \"four board seats. Not three. Because he wants to start with unequivocal control\") are the receipt for the founders' rejection of his terms. The Putin/Xi remarks, the \"$9B of short sale\" security comment, and the AGI-fear \"what the fuck are we gonna do\" line are also Musk's most candid contemporaneous expressions of the AGI-safety motivation that the court has otherwise [[Key Themes|bracketed]]. See also [[PX-1250-52]] (the defense-uploaded Bates page from this same journal volume admitted on Day 7 redirect). For the broader journal arc — including the Aug 21, \"fire Elon,\" and \"morally bankrupt\" passages — see [[Brockman Journal]].\n\n---\n*See also:* [[Day 6|Day 6 digest]] · [[Day 7|Day 7 digest]] · [[PX 151]] · [[PX-157]] · [[PX-1250-52]] · [[Brockman Journal]] · [[Greg Brockman]] · [[Ilya Sutskever]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-156", "exhibit": "PX 156", "party": "Plaintiffs", "type": "Email thread", "admitted_trial_day": "Day 3 (April 29, 2026)", "uploaded_box_pt": "2026-04-29T15:09:19", "uploader": "Someone", "pages": 2, "size_bytes": 173322, "source_pdf": "PX-156.pdf", "bates": "2024MUSK-0009240–0009241", "pdf_url": "https://media.mts-in.com/PX-156.pdf", "body_markdown": "# PX 156 — Sept 13, 2017 Musk \"Re: Current State\" / \"unequivocal control\" email\n\n> Musk's September 13, 2017 7:40 AM reply to Ilya Sutskever's \"Current State\" summary of the proposed for-profit equity, board composition, and credit split: Musk accepts the framework, says he would \"unequivocally have initial control of the company\" with a 7/16 → 25% influence at a 16-person board, and signs off \"I have a really good feeling about this.\"\n\n## Document type\n**Email thread, plain text, two messages.** Top: Musk to Sutskever (cc Brockman), Sep 13, 2017 07:40:05 -0000, Subject \"Re: Current State.\" Bottom (quoted): Sutskever to Musk, Sep 12, 2017 11:44 PM, summarizing equity (\"Greg: 10M grant/10M investment; Sam: 10M grant/10M investment; Ilya: 12M grant/2.5M investment + 5.5M loan from Greg\"), control (\"3 board seats for Elon, 1 board seat each for Ilya/Greg/Sam\"), and credit. Marked Confidential. Two pages. Bates 2024MUSK-0009240–0009241.\n\n## Logistics\n- **Trial admission:** Day 3 (April 29, 2026) — Musk direct concludes; cross begins. Per [[Key Themes]]: \"PX 156 — Sept 13, 2017 Musk email.\" Used by defense on cross to drive the \"unequivocal control\" line of attack on Musk's \"stole-a-charity\" theory.\n- **Box upload:** 2026-04-29 15:09:19 PT — same Day 3 mid-afternoon batch as [[PX 157]], PX 233, PX 295, PX 296, PX 355.\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~173 KB, 2 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `156.pdf`.\n\n## Transcribed text\n\n> **From:** \"Elon Musk\" <erm@spacex.com>\n> **To:** \"Ilya Sutskever\" ▮▮▮▮\n> **Cc:** \"Greg Brockman\" ▮▮▮▮\n> **Subject:** Re: Current State\n> **Date:** Wed, 13 Sep 2017 07:40:05 -0000\n> **Importance:** Normal\n>\n> Sounds good. The three common stock seats (you, Greg and Sam) should be elected by common shareholders. They will de facto be yours, but not in the unlikely event that you lose the faith of a huge percentage of common stockholders over time or step away from the company by choice.\n>\n> I think that the Preferred A investment round (supermajority me) should have the right to appoint four (not three) seats. I would not expect to appoint them immediately, but, like I said I would unequivocally have initial control of the company, but this will change quickly. The rough target would be to get to a 12 person board (probably more like 16 if this board really ends up deciding the fate of the world) where each board member has a deep understanding of technology, at least a basic understanding of AI and strong & sensible morals.\n>\n> Apart from the Series A four and the Common three, there would likely be a board member with each new lead investor/ally. However the specific individual new board members can only be added if all but one existing board members agrees. Same for removing board members.\n>\n> There will also be independent board members we want to add who aren't associated with an investor. Same rules apply: requires all but one of existing directors to add or remove.\n>\n> I'm super tired and don't want to overcomplicate things, but this seems approx right. At the sixteen person board level, we would have 7/16 votes and I'd have a 25% influence, which is my min comfort level. That sounds about right to me. If everyone else we asked to join our board is truly against us, we should probably lose.\n>\n> As mentioned, my experience with boards (assuming they consist of good, smart people) is that they are rational and reasonable. There is basically never a real hardcore battle where an individual board vote is pivotal, so this is almost certainly (sure hope so) going to be a moot point.\n>\n> As a closing note, I've been really impressed with the quality of discussion with you guys on the equity and board stuff. I have a really good feeling about this.\n>\n> Lmk if above seems reasonable.\n>\n> Elon\n\n> **From:** Ilya Sutskever <ilyasu@openai.com>\n> **To:** Elon Musk <erm@spacex.com>\n> **Cc:** Greg Brockman <gdb@openai.com>\n> **Subject:** Current State\n> **Date:** Tue, Sep 12, 2017 at 11:44 PM\n>\n> Hi Elon,\n> >\n> > To summarize our understanding of the current state:\n> >\n> > **On equity:**\n> >\n> > - Greg: 10M grant/10M investment\n> > - Sam: 10M grant/10M investment\n> > - Ilya: 12M grant/2.5M investment + 5.5M loan from Greg securitized by Ilya's YC vested stock, from his work at OpenAI\n> >\n> > **On control:**\n> >\n> > - 3 board seats for Elon, 1 board seat each for Ilya/Greg/Sam. Plan to expand board over time through unanimous consent of current board, up to 12 people. Details TBD on tiebreak and whether the directors are statically allocated.\n> >\n> > **On credit:**\n> >\n> > - Elon is extremely happy to help Greg and Ilya get credit for their work — and is very open to any ideas Greg and Ilya may have.\n> >\n> > Looking forward to the conclusions of your conversation with Sam tomorrow.\n> >\n> > Ilya\n\n## Commentary\n\nPX 156 is the defense's central exhibit for the \"Musk wanted unequivocal control\" theory of the case — see [[Key Themes]] §\"Unequivocal control (the 2017 negotiation breakdown).\" The phrase \"I would unequivocally have initial control of the company\" — Musk's own words, written September 13, 2017 — anchors Savitt's cross of Musk on Day 3: \"So the idea is, you would have unequivocal control of this for-profit when it starts; right?\" — \"Yes.\" This email is also where the *equity* numbers Brockman, Altman, and Sutskever later actually received first appear in discoverable form (\"Greg: 10M grant/10M investment; Sam: 10M grant/10M investment\"), which becomes plaintiffs' counter-anchor for the [[PX 350]] / \"$30 billion / paid $0\" narrative on Day 6 ([[Greg Brockman]]). The exhibit is the textual prelude to the September 20, 2017 \"[[PX 157|Honest Thoughts]]\" email a week later in which Sutskever and Brockman tell Musk to his face that \"during this negotiation, you've shown to us that absolute control is extremely important to you\" — and Musk replies \"I've had enough. This is the final straw.\" Read together, PX 156 and PX 157 are the textual pivot of the whole 2017 control fight.\n\n---\n*See also:* [[Day 3|Day 3 digest]] · [[PX 151]] · [[PX 154]] · [[PX 157]] · [[PX 161]] · [[PX 350]] · [[Brockman Journal]] · [[Greg Brockman]] · [[Ilya Sutskever]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-157", "exhibit": "PX 157", "party": "Plaintiffs", "type": "Email", "admitted_trial_day": "Day 3 (April 29, 2026)", "uploaded_box_pt": "2026-04-29T15:09:19", "uploader": "Someone", "pages": 2, "size_bytes": 401598, "source_pdf": "PX-157.pdf", "pdf_url": "https://media.mts-in.com/PX-157.pdf", "body_markdown": "# PX 157 — Sept 20, 2017 \"final straw\" / \"Honest Thoughts\" email\n\n> Musk's two-paragraph \"final straw\" reply (\"Discussions are over.\") on top of Ilya Sutskever and Greg Brockman's long \"Honest Thoughts\" letter to Musk and Altman — the climactic moment of the August–September 2017 control negotiation.\n\n## Document type\n**Email thread, plain text, two messages.** Musk's Sept 20, 2017 2:17 PM PDT reply on top of Sutskever's Sept 20, 2017 2:08 PM email (sent jointly by Sutskever and Brockman, captioned \"Honest Thoughts\"). Recipients on both messages: Ilya Sutskever, Sam Altman, Greg Brockman, Sam Teller, Shivon Zilis. Bates 2024MUSK-0005100–5101.\n\n## Logistics\n- **Trial admission:** Day 3 (April 29, 2026). Per [[Key Themes]]: \"PX 157 — Sept 20, 2017 final-straw email.\" The climactic moment of plaintiffs' direct examination of Musk and a centerpiece of defense's cross — \"you've shown to us that absolute control is extremely important to you\" (Brockman/Sutskever's words to Musk, quoted in Savitt's opening).\n- **Box upload:** 2026-04-29 15:09:19 PT — Day 3 mid-afternoon batch (clustered with PX 233, PX 296, PX 355).\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~392 KB, 2 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `157.pdf`.\n\n## Transcribed text\n\n> **From:** Elon Musk <erm@spacex.com>\n> **To:** Ilya Sutskever <ilyasu@openai.com>\n> **Cc:** Sam Altman <shga@ycombinator.com>, Greg Brockman <gdb@openai.com>, \"Sam Teller\" <steller@spacex.com>, Shivon Zilis <Shivon@spacex.com>\n> **Subject:** Re: Honest Thoughts\n> **Date:** Wed, 20 Sep 2017 14:17:02 -0700\n>\n> Guys, I've had enough. This is the final straw.\n>\n> Either go do something on your own or continue with OpenAI as a nonprofit. I will no longer fund OpenAI until you have made a firm commitment to stay or I'm just being a fool who is essentially providing free funding for you to create a startup.\n>\n> Discussions are over.\n\n> **From:** Ilya Sutskever <ilyasu@openai.com>\n> **To:** Elon Musk <erm@spacex.com>, Sam Altman <shga@ycombinator.com>\n> **Cc:** Greg Brockman <gdb@openai.com>\n> **Subject:** Honest Thoughts\n> **Date:** Wed, Sep 20, 2017 at 2:08 PM\n>\n> Elon, Sam,\n>\n> This process has been the highest stakes conversation that Greg and I have ever participated in, and if the project succeeds, it'll turn out to have been the highest stakes conversation the world has seen. It's also been a deeply personal conversation for all of us.\n>\n> Yesterday while we were considering making our final commitment given the non-solicit agreement, we realized we'd made a mistake. We have several important concerns that we haven't raised with either of you. We didn't raise them because we were afraid to: we were afraid of harming the relationship, having you think less of us, or losing you as partners.\n>\n> There is some chance that our concerns will prove to be unresolvable. We really hope it's not the case, but we know we will fail for sure if we don't all discuss them now. And we have hope that we can work through them and all continue working together.\n>\n> **Elon:**\n>\n> We *really* want to work with you. We believe that if we join forces, our chance of success in the mission is the greatest. Our upside is the highest. There is no doubt about that. Our desire to work with you is so great that we are happy to give up on the equity, personal control, make ourselves easily firable — whatever it takes to work with you.\n>\n> But we realized that we were careless in our thinking about the implications of control for the world. Because it seemed so hubristic, we have not been seriously considering the implications of success.\n>\n> - The current structure provides you with a path where you end up with unilateral absolute control over the AGI. You stated that you don't want to control the final AGI, but during this negotiation, you've shown to us that absolute control is extremely important to you.\n>   - As an example, you said that you needed to be CEO of the new company so that everyone will know that you are the one who is in charge, even though you also stated that you hate being CEO and would much rather not be CEO.\n> - Thus, we are concerned that as the company makes genuine progress towards AGI, you will choose to retain your absolute control of the company despite current intent to the contrary. We disagree with your statement that our ability to leave is our greatest power, because once the company is actually on track to AGI, the company will be much more important than any individual.\n> - The goal of OpenAI is to make the future good and to avoid an AGI dictatorship. You are concerned that Demis could create an AGI dictatorship. So do we. So it is a bad idea to create a structure where you could become a dictator if you chose to, especially given that we can create some other structure that avoids this possibility.\n>\n> We have a few smaller concerns, but we think it's useful to mention it here:\n>\n> - In the event we decide to buy Cerebras, my strong sense is that it'll be done through Tesla. But why do it this way if we could also do it from within OpenAI? Specifically, the concern is that Tesla has a duty to shareholders to maximize shareholder return, which is not aligned with OpenAI's mission. So the overall result may not end up being optimal for OpenAI.\n> - We believe that OpenAI the non-profit was successful because both you and Sam were in it. Sam acted as a genuine counterbalance to you, which has been extremely fruitful. Greg and I, at least so far, are much worse at being a counterbalance to you. We feel this is evidenced even by this negotiation, where we were ready to sweep the long-term AGI control questions under the rug while Sam stood his ground.\n>\n> **Sam:**\n>\n> When Greg and I are stuck, you've always had an answer that turned out to be deep and correct. You've been thinking about the ways forward on this problem extremely deeply and thoroughly. Greg and I understand technical execution, but we don't know how structure decisions will play out over the next month, year, or five years.\n>\n> But we haven't been able to fully trust your judgements throughout this process, because we don't understand your cost function.\n>\n> - We don't understand why the CEO title is so important to you. Your stated reasons have changed, and it's hard to really understand what's driving it.\n> - Is AGI *truly* your primary motivation? How does it connect to your political goals? How has your thought process changed over time?\n>\n> **Greg and Ilya:**\n>\n> We had a fair share of our own failings during this negotiation, and we'll list some of them here (Elon and Sam, I'm sure you'll have plenty to add…):\n>\n> - During this negotiation, we realized that we have allowed the idea of financial return 2-3 years down the line to drive our decisions. This is why we didn't push on the control — we thought that our equity is good enough, so why worry? But this attitude is wrong, just like the attitude of AI experts who don't think that AI safety is an issue because they don't really believe that they'll build AGI.\n> - We did not speak our full truth during the negotiation. We have our excuses, but it was damaging to the process, and we may lose both Sam and Elon as a result.\n>\n> There's enough baggage here that we think it's very important for us to meet and talk it out. Our collaboration will not succeed if we don't. Can all four of us meet today? If all of us say the truth, and resolve the issues, the company that we'll create will be much more likely to withstand the very strong forces it'll experience.\n>\n> - Greg & Ilya\n\n## Commentary\n\nPX 157 is the central document of the August–September 2017 control fight and the textual hinge between the plaintiffs' and defense's stories of who broke faith with whom. **Plaintiffs' read:** Musk walked away when he didn't get unilateral control, and the founders bent over backward — \"happy to give up on the equity, personal control, make ourselves easily firable\" — to keep the nonprofit on mission. The line **\"I will no longer fund OpenAI\"** is plaintiffs' Exhibit A for Musk treating the donations as conditional, and Brockman's same-day journal entry recording **\"His story will correctly be that we weren't honest with him in the end about still wanting to do the for-profit just without him\"** appears in [[PX 161]]. **Defense's read:** PX 157 is the receipt for \"**unequivocal control**\" — Sutskever and Brockman tell Musk to his face he had been demanding to be CEO, controlling shareholder, and the one \"in charge,\" that he had cited an AGI-dictatorship risk *from himself*, and that the proposed cap-table fight was about him having \"51 percent of the company.\" The defense uses this email to neutralize Musk's \"stole-a-charity\" narrative: the founders said no to him, and Musk picked up his marbles and went home (Cohen/Savitt opening, see [[Key Themes]]). Cross also matched it against Brockman's Aug 21, 2017 \"**This is the only chance we have to get out from Elon**\" passage in [[PX 151]]. The Cerebras concern in this email is the seed for plaintiffs' later Cerebras-conflict-of-interest theory built around Brockman's March 17, 2017 Cerebras stock purchase (see Day 6 testimony in [[Key Themes]]).\n\n---\n*See also:* [[Day 3|Day 3 digest]] · [[PX 151]] · [[PX 154]] · [[PX 161]] · [[Brockman Journal]] · [[Greg Brockman]] · [[Ilya Sutskever]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-158", "exhibit": "PX 158", "party": "Plaintiffs", "type": "Email thread", "admitted_trial_day": "Day 3 (April 29, 2026)", "uploaded_box_pt": "2026-04-29T15:09:20", "uploader": "Someone", "pages": 2, "size_bytes": 404051, "source_pdf": "PX-158.pdf", "pdf_url": "https://media.mts-in.com/PX-158.pdf", "body_markdown": "# PX 158 — Sept 21, 2017 Altman \"i remain enthusiastic about the non-profit structure!\" reply\n\n> Sam Altman's one-line September 21, 2017 reply atop the previous day's [[PX 157|\"final straw\"]] / \"Honest Thoughts\" thread: \"**i remain enthusiastic about the non-profit structure!**\" — the line Musk testified he was \"foolish enough to believe.\"\n\n## Document type\n**Email thread, plain text, three messages stitched.** Altman's September 21, 2017 9:17 AM (UTC) reply on top of Musk's September 20, 2017 3:08 PM \"this is not an ultimatum\" message, on top of the original Sutskever/Brockman \"Honest Thoughts\" letter. Subject \"Re: Honest Thoughts.\" Recipients: Musk, Sutskever, Brockman, Sam Teller, Shivon Zilis. Bates 2024MUSK-0005049–5050.\n\n## Logistics\n- **Trial admission:** Day 3 (April 29, 2026). Pre-trial wiki reference: **\"PX 158 — Sept 21, 2017 Altman reply (nonprofit structure).\"** Used in plaintiffs' Musk direct: Musk testified, \"I understood that what Sam Altman was saying is that OpenAI would remain a nonprofit. And I believed him… So I was foolish enough to believe him.\" `(042926TT.txt:1636)`\n- **Box upload:** 2026-04-29 15:09:20 PT — Day 3 mid-afternoon batch (clustered with [[PX 157]], PX 105, PX 236).\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~395 KB, 2 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `158.pdf`.\n\n## Transcribed text\n\n> **From:** Sam Altman <shga@ycombinator.com>\n> **To:** Elon Musk <erm@spacex.com>, Ilya Sutskever <ilyasu@openai.com>\n> **Cc:** Greg Brockman <gdb@openai.com>, Sam Teller <steller@spacex.com>, Shivon Zilis <Shivon@spacex.com>\n> **Subject:** Re: Honest Thoughts\n> **Date:** Thu, 21 Sep 2017 16:17:40 +0000\n> **Importance:** Normal\n>\n> i remain enthusiastic about the non-profit structure!\n\n> **From:** Elon Musk <erm@spacex.com>\n> **To:** Ilya Sutskever <ilyasu@openai.com>\n> **Cc:** Sam Altman <shga@ycombinator.com>, Greg Brockman <gdb@openai.com>, Sam Teller <steller@spacex.com>, Shivon Zilis <Shivon@spacex.com>\n> **Subject:** Re: Honest Thoughts\n> **Date:** Wed, Sep 20, 2017 at 3:08 PM\n>\n> To be clear, this is not an ultimatum to accept what was discussed before. That is no longer on the table.\n\n> **From:** Ilya Sutskever <ilyasu@openai.com>\n> **To:** Elon Musk <erm@spacex.com>, Sam Altman <shga@ycombinator.com>\n> **Cc:** Greg Brockman <gdb@openai.com>\n> **Subject:** Honest Thoughts\n> **Date:** Wed, Sep 20, 2017 at 2:08 PM\n>\n> Elon, Sam,\n> >\n> > This process has been the highest stakes conversation that Greg and I have ever participated in, and if the project succeeds, it'll turn out to have been the highest stakes conversation the world has seen. It's also been a deeply personal conversation for all of us.\n> >\n> > Yesterday while we were considering making our final commitment given the non-solicit agreement, we realized we'd made a mistake. We have several important concerns that we haven't raised with either of you. We didn't raise them because we were afraid to: we were afraid of harming the relationship, having you think less of us, or losing you as partners.\n> >\n> > There is some chance that our concerns will prove to be unresolvable. We really hope it's not the case, but we know we will fail for sure if we don't all discuss them now. And we have hope that we can work through them and all continue working together.\n> >\n> > **Elon:**\n> >\n> > We *really* want to work with you. We believe that if we join forces, our chance of success in the mission is the greatest. Our upside is the highest. There is no doubt about that. Our desire to work with you is so great that we are happy to give up on the equity, personal control, make ourselves easily firable — whatever it takes to work with you.\n> >\n> > But we realized that we were careless in our thinking about the implications of control for the world. Because it seemed so hubristic, we have not been seriously considering the implications of success.\n> >\n> > - The current structure provides you with a path where you end up with unilateral absolute control over the AGI. You stated that you don't want to control the final AGI, but during this negotiation, you've shown to us that absolute control is extremely important to you.\n> >   - As an example, you said that you needed to be CEO of the new company so that everyone will know that you are the one who is in charge, even though you also stated that you hate being CEO and would much rather not be CEO.\n> > - Thus, we are concerned that as the company makes genuine progress towards AGI, you will choose to retain your absolute control of the company despite current intent to the contrary. We disagree with your statement that our ability to leave is our greatest power, because once the company is actually on track to AGI, the company will be much more important than any individual.\n> > - The goal of OpenAI is to make the future good and to avoid an AGI dictatorship. You are concerned that Demis could create an AGI dictatorship. So do we. So it is a bad idea to create a structure where you could become a dictator if you chose to, especially given that we can create some other structure that avoids this possibility.\n> >\n> > We have a few smaller concerns, but we think it's useful to mention it here:\n> >\n> > - In the event we decide to buy Cerebras, my strong sense is that it'll be done through Tesla. But why do it this way if we could also do it from within OpenAI? Specifically, the concern is that Tesla has a duty to shareholders to maximize shareholder return, which is not aligned with OpenAI's mission. So the overall result may not end up being optimal for OpenAI.\n> > - We believe that OpenAI the non-profit was successful because both you and Sam were in it. Sam acted as a genuine counterbalance to you, which has been extremely fruitful. Greg and I, at least so far, are much worse at being a counterbalance to you. We feel this is evidenced even by this negotiation, where we were ready to sweep the long-term AGI control questions under the rug while Sam stood his ground.\n> >\n> > **Sam:** [section continues — see [[PX 157]] for full Sutskever/Brockman text]\n> >\n> > **Greg and Ilya:** [closing reflections — see [[PX 157]]]\n> >\n> > - Greg & Ilya\n\n## Commentary\n\nPX 158 is the day-after pivot from the [[PX 157|\"final straw\"]] thread: after Musk's \"Discussions are over\" ultimatum, Altman closes the loop with a single declarative sentence — **\"i remain enthusiastic about the non-profit structure!\"** Plaintiffs build their misrepresentation theory on this line: Musk testified he relied on it to keep funding OpenAI as a nonprofit through 2018 (see [[Key Themes#\"I was a fool\"]]). Defense's read is the opposite: Altman *kept* the nonprofit structure (the LP didn't launch until August 2018), and Musk's complaint is that he didn't get the for-profit *with himself in control*. Musk's intervening \"this is not an ultimatum\" message is rhetorically interesting — it walks back the previous day's \"final straw\" while keeping the funding cutoff in place. Cross-reference: this email is the bookend to [[PX 157]] and the predicate for [[PX 161|Brockman's same-week journal entry]] (\"his story will correctly be that we weren't honest with him in the end about still wanting to do the for-profit just without him\").\n\n---\n*See also:* [[Day 3|Day 3 digest]] · [[PX 157]] · [[PX 151]] · [[PX 161]] · [[Sam Altman]] · [[Brockman Journal]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-159", "exhibit": "PX 159", "party": "Plaintiffs", "type": "Email thread", "admitted_trial_day": "Day 3 (April 29, 2026)", "uploaded_box_pt": "2026-04-29T15:09:20", "uploader": "Someone", "pages": 2, "size_bytes": 247574, "source_pdf": "PX-159.pdf", "pdf_url": "https://media.mts-in.com/PX-159.pdf", "body_markdown": "# PX 159 — Sept 22, 2017 Zilis → Musk debrief, \"Sam threw out a $100M figure\"\n\n> Shivon Zilis's two-day-after-the-final-straw debrief to Musk reporting back from Altman: Altman \"great with keeping non-profit,\" \"lost a lot of trust with Greg and Ilya,\" needs a 10-day hiatus to think, and \"threw out a $100M figure\" he might personally contribute \"if OpenAI stayed a non-profit.\"\n\n## Document type\n**Email thread, three messages.** Friday Sept 22, 2017 5:54 PM PT — Zilis to Musk (cc Sam Teller), structured under five bold headings (Structure / Trust / Hiatus / Fundraising / Communications), on top of a same-morning back-and-forth between Zilis and Musk. Bates 2024MUSK-0010384–85.\n\n## Logistics\n- **Trial admission:** Day 3 (April 29, 2026) — used during Musk direct (concluding) and as setup for the cross. Referenced in [[Key Themes]]: \"PX 159 — referenced in Day 2-3.\"\n- **Box upload:** 2026-04-29 15:09:20 PT — Day-3 mid-afternoon batch (clustered with [[PX-157]], PX 233, PX 296, PX 355).\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~242 KB, 2 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `159.pdf`.\n\n## Transcribed text\n\n> **From:** Shivon Zilis <Shivon@spacex.com>\n> **To:** Elon Musk <erm@spacex.com>\n> **Cc:** Sam Teller <steller@spacex.com>\n> **Subject:** Re: Non-profit\n> **Date:** Friday, September 22, 2017 5:54:23 PM\n>\n> From Altman:\n>\n> **Structure:** Great with keeping non-profit and continuing to support it.\n>\n> **Trust:** Admitted that he lost a lot of trust with Greg and Ilya through this process. Felt their messaging was inconsistent and felt childish at times.\n>\n> **Hiatus:** Sam told Greg and Ilya he needs to step away for 10 days to think. Needs to figure out how much he can trust them and how much he wants to work with them. Said he will come back after that and figure out how much time he wants to spend.\n>\n> **Fundraising:** Greg and Ilya have the belief that 100's of millions can be achieved with donations if there is a definitive effort. Sam thinks there is a definite path to 10's of millions but TBD on more. He did mention that Holden was irked by the move to for-profit and potentially offered more substantial amount of money if OpenAI stayed a non-profit, but hasn't firmly committed. **Sam threw out a $100M figure for this if it were to happen.**\n>\n> **Communications:** Sam was bothered by how much Greg and Ilya keep the whole team in the loop with happenings as the process unfolded. Felt like it distracted the team. On the other hand, apparently in the last day almost everyone has been told that the for-profit structure is not happening and he is happy about this at least since he just wants the team to be heads down again.\n>\n> Shivon\n\n> **From:** Elon Musk <erm@spacex.com>\n> **To:** Shivon Zilis <Shivon@spacex.com>\n> **Subject:** Re: Non-profit\n> **Date:** Fri, Sep 22, 2017 at 10:01 AM\n>\n> Ok\n\n> **From:** Shivon Zilis <Shivon@spacex.com>\n> **To:** Elon Musk <erm@spacex.com>\n> **Subject:** Non-profit\n> **Date:** Fri, Sep 22, 2017 at 9:50 AM\n>\n> Hi Elon,\n> >>\n> >> Quick FYI that Greg and Ilya said they would like to continue with the non-profit structure. They know they would need to provide a guarantee that they won't go off doing something else to make it work.\n> >>\n> >> Haven't spoken to Altman yet but he asked to talk this afternoon so will report anything I hear back.\n> >>\n> >> If anything I can do to help let me know.\n\n## Commentary\n\nPX 159 is plaintiffs' best contemporaneous evidence that **two days after the [[PX-157|final-straw email]], Sam Altman was telling Zilis (Musk's information conduit) that the for-profit was off, the nonprofit was on, that Greg and Ilya had behaved \"childishly,\" and that he himself would commit \"$100M\" if OpenAI stayed a nonprofit.** That last representation is the quiet bombshell of the exhibit — and the foil for plaintiffs' Day 6/7 evidence ([[Brockman Journal]] PX 161, PX 163) that Brockman in November 2017 was already privately writing that \"we should just flip to a full profit. Making money for us sounds great.\" The \"Holden was irked\" reference is to Holden Karnofsky (OpenPhil), who in early 2017 had committed $30M to OpenAI as a nonprofit — a separate restraint on the for-profit pivot that the founders were navigating. Cross-reference Zilis's August 2017 contemporaneous \"ironclad agreement to not have Elon or anyone have absolute control of AGI\" note (Day 8 testimony — see [[Key Themes]] §\"Day 8 — what Zilis added\").\n\n---\n*See also:* [[Day 3|Day 3 digest]] · [[PX-157]] · [[Shivon Zilis]] · [[Sam Altman]] · [[Greg Brockman]] · [[Brockman Journal]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-16", "exhibit": "PX 16", "party": "Plaintiffs", "type": "Articles of incorporation", "admitted_trial_day": "Day 2 (April 28, 2026)", "uploaded_box_pt": "2026-04-28T14:51:59", "uploader": "Someone", "pages": 6, "size_bytes": 700606, "source_pdf": "PX-16.pdf", "pdf_url": "https://media.mts-in.com/PX-16.pdf", "body_markdown": "# PX 16 — OpenAI, Inc. Delaware Certificate of Incorporation (2015) and Amended & Restated Certificate (2020)\n\n> Delaware Secretary of State certified packet of OpenAI, Inc.'s original Dec. 8, 2015 nonprofit Certificate of Incorporation and its Apr. 23, 2020 Amended & Restated Certificate — the foundational corporate-law document of plaintiffs' charitable-trust theory.\n\n## Document type\n**Articles of incorporation (certified Delaware corporate filings).** Two stacked certificates: (1) the Dec. 8, 2015 Certificate of Incorporation of OpenAI, Inc. as a nonprofit non-stock corporation, signed by incorporator Jonathan Levy at 335 Pioneer Way, Mountain View, CA; and (2) the Apr. 23, 2020 Amended & Restated Certificate executed by OpenAI Secretary Chris Clark. Authenticated by Delaware Secretary of State Jeffrey W. Bullock on March 9, 2022.\n\n## Logistics\n- **Trial admission:** Day 2 (April 28, 2026) — openings + Musk direct begins.\n- **Box upload:** 2026-04-28 14:51:59 PT.\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~684 KB, 6 pages. Bates OPENAI_MUSK00000416–00000421.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `016.pdf`.\n\n## Transcribed text\n\n**[Page 1 — Delaware Secretary of State authentication]**\n\n> Delaware\n> The First State\n>\n> I, JEFFREY W. BULLOCK, SECRETARY OF STATE OF THE STATE OF DELAWARE, DO HEREBY CERTIFY THE ATTACHED ARE TRUE AND CORRECT COPIES OF ALL DOCUMENTS ON FILE OF \"OPENAI, INC.\" AS RECEIVED AND FILED IN THIS OFFICE.\n>\n> THE FOLLOWING DOCUMENTS HAVE BEEN CERTIFIED:\n> CERTIFICATE OF INCORPORATION, FILED THE EIGHTH DAY OF DECEMBER, A.D. 2015, AT 2:22 O'CLOCK P.M.\n> RESTATED CERTIFICATE, FILED THE TWENTY-THIRD DAY OF APRIL, A.D. 2020, AT 7:05 O'CLOCK P.M.\n>\n> AND I DO HEREBY FURTHER CERTIFY THAT THE AFORESAID CERTIFICATES ARE THE ONLY CERTIFICATES ON RECORD OF THE AFORESAID CORPORATION, \"OPENAI, INC.\".\n>\n> 5902936 8100H — SR# 20220936493\n> Authentication: 202868248 — Date: 03-09-22\n\n**[Pages 2–3 — December 8, 2015 Certificate of Incorporation]**\n\n> CERTIFICATE OF INCORPORATION OF A NON-STOCK CORPORATION\n> OPENAI, INC.\n>\n> *State of Delaware, Secretary of State, Division of Corporations — Delivered 02:22 PM 12/08/2015 — FILED 02:22 PM 12/08/2015 — SR 20151247198 — File Number 5902936*\n>\n> **FIRST:** The name of the Corporation is \"OpenAI, Inc.\" (the \"Corporation\").\n>\n> **SECOND:** The address of the Corporation's registered office in the State of Delaware is 2711 Centerville Road, Suite 400, Wilmington, New Castle County, Delaware 19808. The name of its registered agent at such address is Corporation Service Company.\n>\n> **THIRD:** This Corporation shall be a nonprofit corporation organized exclusively for charitable and/or educational purposes within the meaning of section 501(c)(3) of the Internal Revenue Code of 1986, as amended, or the corresponding provision of any future United States Internal Revenue law. **The specific purpose of this corporation is to provide funding for research, development and distribution of technology related to artificial intelligence. The resulting technology will benefit the public and the corporation will seek to open source technology for the public benefit when applicable. The corporation is not organized for the private gain of any person.** In furtherance of its purposes, the corporation shall engage in any lawful act of activity for which nonprofit corporations may be organized under the General Corporation Law of Delaware.\n>\n> **FOURTH:** This corporation is organized and operated exclusively for purposes set forth in Article THIRD hereof within the meaning of the Internal Revenue Code section 501(c)(3). No substantial part of the activities of this corporation shall consist of carrying on propaganda, or otherwise attempting to influence legislation and this corporation shall not participate or intervene in any political campaign... Notwithstanding any other provision of these articles, the corporation shall not carry on any other activities not permitted to be carried on (a) by a corporation/organization exempt from Federal income tax under section 501(c)(3)... or (b) by a corporation/organization, contributions to which are deductible under section 170(c)(2) of the Internal Revenue Code...\n>\n> **FIFTH:** The property of this corporation is **irrevocably dedicated** to the purposes in Article THREE hereof and **no part of the net income or assets of this corporation shall ever inure to the benefit of any director, officer or member thereof or to the benefit of any private person.** Upon the dissolution or winding up of this corporation, its assets remaining after payment, or provision for payment, of all debts and liabilities of this corporation shall be distributed to a nonprofit fund, foundation, or corporation which is organized and operated exclusively for charitable, educational and/or religious purposes and which has established its tax exempt status under Internal Revenue Code section 501(c)(3)...\n>\n> **SIXTH:** The corporation shall not have any capital stock.\n>\n> **SEVENTH:** The corporation shall not have any members.\n>\n> **EIGHTH:** The name and mailing address of the incorporator are as follows:\n> Jonathan Levy\n> 335 Pioneer Way\n> Mountain View, CA 94041\n>\n> I, Jonathan Levy, for the purpose of forming a corporation under the laws of the State of Delaware, do make, file and record this Certificate, and do certify that the facts herein stated are true...\n>\n> Executed on December 8, 2015.\n> AUTHORIZED OFFICER\n> /s/ Jonathan Levy\n\n**[Pages 4–6 — April 23, 2020 Amended and Restated Certificate of Incorporation]**\n\n> AMENDED AND RESTATED CERTIFICATE OF INCORPORATION OF OPENAI, INC.\n> A NONPROFIT NON-STOCK CORPORATION\n>\n> *State of Delaware — Delivered 07:05 PM 04/23/2020 — FILED 07:05 PM 04/23/2020 — SR 20203114775 — File Number 5902936*\n>\n> Pursuant to Sections 242 and 245 of the General Corporation Law of the State of Delaware, OpenAI, Inc., a nonprofit non-stock corporation organized and existing under the laws of the State of Delaware, and originally incorporated under the same name on December 8, 2015, does hereby certify:\n>\n> 1. That the Certificate of Incorporation of this corporation is hereby Amended and Restated as set forth in the attached Amended and Restated Certificate of Incorporation.\n> 2. That the attached Amended and Restated Certificate of Incorporation was duly adopted by the Board of Directors in accordance with the provisions of Section 242 of the General Corporation Law of the State of Delaware.\n>\n> IN WITNESS WHEREOF, OpenAI, Inc. has caused this Certificate to be executed by Chris Clark, its authorized officer, this 14th day of January, 2020.\n>\n> By: /s/ Chris Clark\n> Name: Chris Clark\n> Title: Secretary\n>\n> ---\n>\n> CERTIFICATE OF INCORPORATION OF A NON-STOCK CORPORATION\n> OPENAI, INC.\n>\n> **FIRST:** The name of the Corporation is \"OpenAI, Inc.\" (the \"Corporation\").\n>\n> **SECOND:** The address of the Corporation's registered office in the State of Delaware is 251 Little Falls Drive, Wilmington, New Castle County, Delaware 19808. The name of its registered agent at such address is Corporation Service Company.\n>\n> **THIRD:** This Corporation shall be a nonprofit corporation organized exclusively for charitable purposes within the meaning of section 501(c)(3) of the Internal Revenue Code of 1986, as amended, or the corresponding provision of any future United States Internal Revenue law. **The specific purpose of this corporation is to ensure that artificial general intelligence benefits all of humanity, including by conducting and/or funding artificial intelligence research.** The corporation may also research and/or otherwise support efforts to safely develop and distribute such technology and its associated benefits, including analyzing the societal impacts of the technology and supporting related educational, economic, and safety policy research and initiatives. The resulting technology will benefit the public and **the corporation will seek to distribute it for the public benefit when applicable.** The corporation is not organized for the private gain of any person...\n>\n> **FOURTH:** [501(c)(3) operating restrictions — substantively identical to 2015 version.]\n>\n> **FIFTH:** The property of this corporation is **irrevocably dedicated** to the purposes in Article THIRD hereof and no part of the net income or assets of this corporation shall ever inure to the benefit of any director, officer or member thereof or to the benefit of any private person...\n>\n> **SIXTH:** The corporation shall not have any capital stock.\n>\n> **SEVENTH:** **The corporation shall have one or more members, and the conditions of membership shall be stated in the Bylaws.** [Note: This reverses the 2015 \"shall not have any members\" provision.]\n\n## Commentary\n\nPX 16 is the corporate-law spine of plaintiffs' [[Key Themes|\"stole a charity\"]] theory: the 2015 charter says OpenAI's purpose is \"**research, development and distribution of technology related to artificial intelligence**\" and that the corporation \"**will seek to open source technology for the public benefit**\" — language plaintiffs use to anchor Musk's \"**The 'OPEN' in OpenAI represents open source**\" testimony on Day 2 and the judge's pointed admonition to defense counsel about the [[PX-152|PTO trademark prosecution history]]. The 2020 Amended & Restated Certificate is equally important to the defense: the purpose clause was rewritten from \"fund AI research\" to \"**ensure that artificial general intelligence benefits all of humanity**,\" the open-source commitment was softened to merely \"distribute it for the public benefit when applicable,\" and Article SEVENTH was flipped to permit members — corporate-law scaffolding for the LP/capped-profit structure. Both versions retain the key Article FIFTH irrevocable-dedication clause that plaintiffs treat as the charitable-trust anchor under California Corp. Code § 5142 and § 5250. See [[Day 2|Day 2 digest]] for openings; the 2020 restated charter is the same vintage as [[PX-103]] (the July 2020 rent solicitation) which plaintiffs use to defeat the statute of limitations.\n\n---\n*See also:* [[Day 2|Day 2 digest]] · [[PX-152]] · [[PX-103]] · [[Key Themes]] · [[Case Overview]]\n"}
{"exhibit_id": "PX-161", "exhibit": "PX 161", "party": "Plaintiffs", "type": "Personal journal entry", "admitted_trial_day": "Day 6 (May 4, 2026)", "uploaded_box_pt": "2026-05-04T14:46:10", "uploader": "Someone", "pages": 10, "size_bytes": 470899, "source_pdf": "PX-161.pdf", "pdf_url": "https://media.mts-in.com/PX-161.pdf", "body_markdown": "# PX 161 — Nov 6, 2017 Brockman journal entry (\"morally bankrupt\" / \"lie\" / \"his story will correctly be\")\n\n> Greg Brockman's ten-page contemporaneous laptop-journal entry from November 6, 2017 — the longest, most quoted entry in the trial. Captures the morning-after of a Sutskever–Brockman conclusion that they \"morally should not be kicking elon out,\" the pre-meeting scripts with Sam Altman, the post-meeting \"shit, i feel so good,\" the \"**morally bankrupt**\" reflection on what it would mean to convert to a B-corp without Musk, and the \"**if three months later we're doing b corp then it was a lie**\" passage about what they could and couldn't tell Musk.\n\n## Document type\n**Personal journal entry, plain ASCII text, ten pages.** Filename embedded in document: `2017-11-06.txt`. Stream-of-consciousness diary kept on Brockman's laptop since 2010 (see [[Brockman Journal]]). Bates OPENAI_MUSK00039470–39479; stamped \"Highly Confidential.\" Several large black-box \"Redacted\" boxes (one full-page) for personal/private content. Includes embedded Signal-style messages from Shivon Zilis and Greg Brockman at the close.\n\n## Logistics\n- **Trial admission:** Day 6 (May 4, 2026). The entry plaintiffs returned to most often. Used by **Mr. Kry** on cross-examination of Greg Brockman, then re-read by Mr. Kry on Day 7 redirect to put each line back in chronological context.\n- **Box upload:** 2026-05-04 14:46:10 PT — Day 6 mid-afternoon batch (clustered with PX 151, PX 241, PX 390).\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~460 KB, 10 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `161.pdf`.\n\n## Transcribed text\n\n> **File Name: 2017-11-06.txt**\n>\n> Next day: Ilya feeling like **we morally should not be kicking elon out**, and should be trying to make the non-profit work and convince him to stay.\n>\n> Going to do 9a call with Sam. On it the questions are:\n>\n> 1. Suppose we do non-profit and E stays, are you cool with us doing hardware in tesla?\n> 2. we cou\n>\n> Message to E could be:\n> - more money won't move us faster right now.\n> - will start some fundraising convos now, ship the hand, and see what fundraising looks like after that.\n> - we can make this work. if you have a competitor then we're in trouble.\n> - in the meanwhile we'll help jim do the hardware right.\n>\n> [Call notes:]\n> - sam: say, we wanna do openai, if you wanna do inside tesla it's ok. if weirdly competitive should revisit.\n> - ilya: was thinking, what we need.\n> - sam: depends how strategically dependent we would be, could imagine a way that works.\n> - ilya: earlier more concerned. don't want only strategy to be tesla. wouldn't wanna be locked into something that only tesla\n> - gdb: why be ok with similar effort in tesla?\n> - sam: would like E to stay on board, not go to war. don't think will work, tell him that. maybe he'll get the hardware built inside of tesla.\n> - ilya: if i'm truly honest, my prefs here are not set in stone, some pref for b-corp. can be made happy in the non-profit world.\n> - sam: would prefer to have him involved.\n> - sam: much more interested in you two being motivated + non-distracted. normally have one thing. that's moved around a lot. thing i care about the most optimum output. if you guys are not gonna be happy with the non-profit, let's figure out the b-corp. if you don't want e involved,\n> - do have weakly held views, but pale in comparison to making you guys fully happy.\n>\n> ** ilya and me breakfast\n> - conclusion is we truly want the b-corp. honestly we also want to get back to work. but it's not super clear how we get there.\n> - one realization is that we always have the option to quit and put up the public job posting. also we're *so* close to the hand (and probably the 5v5) that we should finish those up for sure.\n> - one sad outcome here would be to \"wind down\" the non-profit on those. hopefully not going to be so needed; would only really be needed because elon blocks us going b-corp.\n>\n> two reasons we don't wanna do tesla:\n> - happy to not become rich on this, so long as no one else is. don't see the economics working out.\n> - don't see story for how the tech ends up being owned by the world\n>\n> so our statement to sam: we truly prefer the b-corp. we want to make it happen, though there are costs and difficulties which would stop us. but we think that makes the most sense.\n>\n> if he starts the competitor going to be less\n>\n> **Cannot say that we are committed to the non-profit. don't wanna say that we're committed. if three months later we're doing b corp then it was a lie.**\n>\n> statement must be, right now we have these concerns, we want to see the fundraising landscape. hypothesis: after robot hand and 5v5, totally diff world. we need, upper bound, $150M for 2018. and right now more dollars wouldn't make us move faster.\n>\n> and so... what does this all mean. we are really unsure if he starts a competitor that it's going to end in a good place. btw right now we have this great position of power. but it can of course all go away very quickly.\n>\n> - policy gradients\n>\n> ok, so not feeling so great about all of this. **the true answer is that we want him out.**\n>\n> ok, so talking to shivon: she's like how are you gonna raise the billions in the non-profit? of course E isn't going to put personal money on the line, not going to give up on mars.\n>\n> if he really thinks that AGI is gonna happen first and be the craziest thing and change everything, then why not?\n>\n> can't see us turning this into a for-profit without a very nasty fight. i'm just thinking about the office and we're in the office. and **his story will correctly be that we weren't honest with him in the end about still wanting to do the for profit just without him.**\n>\n> ok so if we're gonna quit would very much like to do it from the position of strength, shipping hand, 5v5, maybe announcing that policy gradients can do all this credit assignment and isn't it great.\n>\n> so what's the outcome we most prefer here? is it, he's out? guess i don't know how it'll end. is it that we both retain the non-profit and then ilya and i quit and start another thing?\n>\n> ** compute\n>\n> so we have this whole GCE deal lined up which could be great. $50M a year to get 400k cores and 5k GPUs. would really put us in business in a way we are not currently.\n>\n> $50M/yr is a lot of money. Or $60M for one year. and people really wanna use the GCP. szymon was practically drooling over it.\n>\n> geez, elon would think it was the true betrayal for us to start getting sponsorship from google. flip side is we kinda need to do it. talking to ilya maybe not.\n>\n> ok, so the conclusion is that we can say:\n> 1. we don't think the full fundraise has been done in the non-profit. especially post the hand, we think it'll be a different game.\n> 2. happy to not become rich on this, so long as no one else is. don't see the economics working out.\n> 3. structure is definitely less \"pure\". that's not a deal breaker for us, but it makes us less excited.\n>\n> we do like being in charge, but i guess it's less of a primary thing for us? we vie\n>\n> don't see story for how the tech ends up being owned by the world\n>\n> ** thinking more\n> can we get the $50M by 2 weeks? i really don't see it happening. greg: $10m. reid: $10m. i guess we could shave the price down further potentially. especially given the azure deal.\n>\n> [...]\n>\n> **ordered prefs:**\n> - b-corp w/ appropriate control structure\n> - non-profit w/ lots of funding\n> - us quitting and doing our own thing\n>\n> we have a preference for the b-corp, would prefer not to tell elon we lurve the non-profit.\n>\n> ** 7p before the meeting\n> ok, so we can say that for our previous meeting, we were ready to commit to the non-profit, given enough funding. it's not that\n>\n> we don't think it's reasonable for us to ask for huge compensation in tesla, and in the non-profit we're all happy to forgo large comp together. doesn't seem like deciding now vs later matters much, and would prefer to get the big results in the non-profit and try to fundraise.\n>\n> - it is a large number, but it's risky stock. [Redacted - Personal] [Redacted - Personal]\n>\n> **if he asks, ok are you going to commit, we'd say, well, we're not sure that the raise can happen. so that's our holdback. we believe in the mission, and always have. but right now there are so many factors and we're so close, would like to just try to get the big results. otherwise openai is a bit failure.**\n>\n> teller:\n> spoke to teller / 21 minutes ago\n> he said elon tried to solve for the thing that was most important for us, which is the control / 21 minutes ago\n> i told him that tesla is already building the hardware / 20 minutes ago\n> he said, elon is very happy with the nonprofit, the disagreement is about what gets moved to tesla / 20 minutes ago\n> anyway, gonna be fun / 20 minutes ago\n> oh ho 20 minutes ago\n> so elon was thinking about us / 20 minutes ago\n> and the cost function that we've communicated\n>\n> [Redacted (full-page block)]\n>\n> ** post meeting\n>\n> **shit, i feel so good.**\n>\n> the meeting was simple. we said that we want to get more results in the non-profit and to fundraise there. he asked if there have been any new commitments, i talked about the azure + GCP deals and mentioned sam's been talking to reid, etc. he was like, azure's not going to just give you $280M. would be way too massive. i said yeah, can see it. last deal was a lot.\n>\n> we also showed him our deck (and i mentioned that we'd been putting these together over the weekend, and wanted to show them to him today. he said yeah what was up with that email? sounded like you were being deceptive. i said, just an accidental BCC, and then we were going to have this meeting today and wanted you to see them here. he said, you should have replied with context and sharing the document, so it doesn't look like you're both boneheaded and dumb.)\n>\n> he was very engaged with the deck, all 40 slides of it.\n>\n> he was like, ok. fundraising is super hard. i've worked hard on gates and he didn't even stop by office. dustin donated but doesn't even show up personally. that's how much these people care about this. i didn't expect to be 70% of the cash contributions. thought i'd be one piece. there are all these names on the website and they haven't contributed. reality check, no one's gonna give you this amount of money, when they see the $10B they're going to run to the hills.\n>\n> one of the two of us is not based in reality.\n>\n> but he was super supportive of us trying, and said, go ahead, i'm happy with anyone (even people who just want to get access to me), happy to give them any fancy title, diminish my own title, etc.. if you can do it then i want to learn from you. i'm very good at getting people to part with their money, i could raise $1b for tesla in a week and have you seen that company's financials?\n>\n> anyway, definitely the best this meeting could have gone. we now have a focus, a goal, and if we accomplish it, then we'll really have shown him that we can outperform him even at something he's great at. if we fail, well, we'll deal with it then.\n>\n> btw another realization from this is that **it'd be wrong to steal the non-profit from him. to convert to a b-corp without him. that'd be pretty morally bankrupt. and he's really not an idiot.**\n>\n> [Embedded Signal exchange:]\n> Greg Brockman: btw thinking of sending this to e and teller, think they'd appreciate or weird?\n> [Redacted - Personal]\n> – gdb\n> Shivon Zilis: Hahaha not weird but i guarantee they are booked so up to you!\n> Greg Brockman: heh, thanks. Think I'll send anyway\n\n## Commentary\n\nPX 161 is the centerpiece of the entire two-day fight over Brockman's journal. Every side of the case turns on a different line of this entry. Mr. Kry's Day 6 cross opened with \"Cannot say that we are committed to the non-profit … if three months later we're doing b corp then it was a lie\" — putting the question to Brockman: if you knew on November 6, 2017 that telling Musk \"we believe in the mission\" while privately preferring B-corp would be a lie, and you said it anyway, what did that make the September 2017 representations to Musk that you were \"happy to give up on the equity\" (see [[PX-157]])? Brockman's defense read: **\"morally bankrupt\"** referred narrowly to *voting Musk off the board and creating a B Corp without him* — not to a for-profit conversion as such — and is corroborated by Brockman's earlier same-week journal entry [[DX-1252]] introduced by defense as a prior consistent statement (\"Real decision is fire Elon\" / \"So we seem converged on the Fire Elon route\"). Mr. Molo's Day 7 recross frame: \"It's just an after-the-fact manufactured excuse for the very honest emotion that you expressed in the journal at the time.\" The \"his story will correctly be that we weren't honest with him in the end\" line is plaintiffs' single most damaging quotation in the case — Brockman's own contemporaneous concession that **the founders were being dishonest with Musk in November 2017**. The \"shit, i feel so good\" / 70%-of-the-cash-contributions / \"$10B\"-fundraise / Bill Gates / Dustin Moskovitz block is the post-meeting color that places Musk and Brockman in the room arguing fundraising realism. See [[Brockman Journal]] for the full sequence of journal passages and [[Day 6|Day 6]] / [[Day 7|Day 7]] digests for the in-court quotations.\n\n---\n*See also:* [[Brockman Journal]] · [[Day 6|Day 6 digest]] · [[Day 7|Day 7 digest]] · [[PX-151]] · [[PX-157]] · [[Greg Brockman]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-163", "exhibit": "PX 163", "party": "Plaintiffs", "type": "Personal journal entry", "admitted_trial_day": "Day 6 (May 4, 2026)", "uploaded_box_pt": "2026-05-04T14:46:09", "uploader": "Someone", "pages": 9, "size_bytes": 406129, "source_pdf": "PX-163.pdf", "pdf_url": "https://media.mts-in.com/PX-163.pdf", "body_markdown": "# PX 163 — Brockman journal, November 12, 2017 (\"flip to a for-profit\")\n\n> Brockman's contemporaneous laptop journal entry six days after the Nov 6, 2017 in-person meeting where he told Musk he wanted to keep fundraising for the nonprofit — written in Brockman's stream-of-consciousness style and containing the now-famous \"flip to a for profit\" and \"Being the Kings of AI is not so bad\" passages.\n\n## Document type\n**Personal journal entry**, file name `2017-11-12.txt` per the document header. Stream-of-consciousness laptop diary kept on Brockman's laptop since 2010, with embedded Signal/IM messages from a colleague (timestamps \"X minutes ago\" interleaved with Brockman's prose). Heavily redacted (\"Redacted - Personal\" boxes). Bates OPENAI_MUSK00039484–00039492.\n\n## Logistics\n- **Trial admission:** Day 6 (May 4, 2026). Used by **Mr. Kry** in cross-examination of Greg Brockman. Per the wiki reference index: \"PX 163 — Brockman journal Nov 12, 2017 (flip to a full profit).\"\n- **Box upload:** 2026-05-04 14:46:09 PT — uploaded the same Day 6 morning the cross began.\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~397 KB, 9 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `163.pdf`.\n\n## Transcribed text\n\n> File Name: 2017-11-12.txt\n\n> ** our plan\n>\n> ok, so the thing to think through is what path we really want to take. it would be nice to be making the billions and to have a concrete technical problem we can go and solve and all that.\n>\n> [Redacted - Personal]\n>\n> we've been thinking that maybe we should just flip to a for profit. making the money for us sounds great and all. [Redacted - Personal]\n>\n> the other side of the coin is that maybe we can work out sometihng with companies like quora. ask for a bounty on problems. if we solve them you get the exclusive access to the trade secrets. now why would people want to donate to fund that?\n>\n> - ligo: $250M gamble\n> - media lab: $45M annual budget (darragh also says has bad reputation)\n> - cern: $1.2B annual budget (that's more like it)\n>\n> the weird thing here is that companies are happy to put in the R&D effort. so our point isn't making the discoveries, it's ensuring that it's a public good.\n>\n> - reid hasn't replied\n> - dustin doesn't show up\n>\n> pretty clear no one is acting as if they really believe or care.\n>\n> the adam model, does it give us the short term cash? no it doesn't. but it maybe causes people to join the coalition.\n>\n> [Redacted]\n>\n> why should people part with their hard-earned money? reid hasn't actually given yet, nor greg jensen. dustin doesn't even show up. elon puts the time, sam puts a small amount of time.\n>\n> we were a bit too rushed on the for-profit transition before, i think. fundamentally right now we haven't found the thing to be high conviction on. and people are being whiplashed by that.\n>\n> but what do we think now? we just want something that'll get us the money. we don't have a plan for productization, but actually by working with companies and asking them for problems which are on critical path we could be in good shape.\n>\n> it all comes down to the money. we can get it from tesla, probably. we could also probably get it from google. maybe we can get it while staying independent and having this consortium of companies that will put the money in exchange for no particular thing except not being left behind.\n>\n> ilya: one possible path to a solution: if coalition happens, then have diff countries with powerful AGIs.\n>\n> the only option:\n> - coalition\n> - tesla\n> - quit\n>\n> ** ilya\n> - try not to just read off the slides\n> - dota code\n> - 7-14 days: show some datapoints, other points of science. as stated on the slide, feels like assertion\n> - make the argument about the niche:\n> - maybe dota bot (confidential) bit earlier too\n> - talk a bit more about the dumb human/einstein gap\n> - solving accident risk: clearer that no one else cares?\n> - agi IP resizing\n>\n> - john notes: at a super high level, the pitch is too long. most meeting slots are an hour. good thing to shoot for 20-30 mins of presentation.\n> - highest level thing: seen startup pitches, all good pitches follow: narrative Q&A format. startups start with broad statement: we are building an app for sharing photos with your friends. really good ones follow path: when receiving \"how does it work?\" \"you're probably wondering\"...\n> - pitch structured logical: here's what we've done, here's what we're doing next. instead: we are creating a non-profit that's developing human level intelligence. why is that a good thing for the world? as opposed by the info that we have to convey.\n> - 1. short + tighten\n> - 2. structure as start with assertion and assertion, questions as someone considering investing might have.\n> - if considering investing in this, qs that didn't get treatment: is this good for the world? sort of addressed, can paint a pretty good picture: world of abundance is good, spend a bunch of time on safety, a box that needs checked, but less time on why this is a thing that you should want to have exist. we're in the world of openai. ford foundation: money to openai or openphil, or to bill + melinda. that's where they are.\n> - tactical stuff: framing of defining AGI economically useful kinda interesting. don't worry about poetry better than humans.\n>\n> [Redacted]\n>\n> - slide design: sat through a lot of presentations. sometimes too text heavy, slides should be duplicative. animations visually distracting. concentrate on deep intellectual point. if do it with build once. talking through it. play once.\n> - ilya: make it more interactive, break up the flow. lot of lecture. engage the audience, vary the format. speakers do dumb tricks like vary tone of voice. switching between the two of us.\n> - something same code as dota, deep and interesting point. could do something with that.\n> - another question: as i watch, success feels very binary, i'll give people some money, then replaced by robot or i won't, laying out some amount of funding milestones. can see what success will look like. throughout the presentation, feels binary outcome. potential donor, map of success in this space.\n>\n> [Redacted]\n>\n> - not sure all the ownership stuff, coalition, helps. kinda needed if companies on board to donate. pitch not to the companies. feels sci-fi to me. up to that point, if you made the pitch that this accelerates economic, next 10x of improvement, drug discovery, cure the illnesses, basic to enumerate, foundation people super stupid approach to cancer.\n> - would talk about the openness. helps, intuitive idea: science is all about scientific commons, making this work avail to benefit all. instinct of googles commercialize, optimize ad words. for us it's good goals for the world. without reaching takeoff, next level up in human cog ability. as you get into the details + levels, feels cabalistic, ford foundation. thought i was curing cancer, didn't realize. don't raise -> don't think too hard. fact that we're a non-profit. 501c3 granted magical powers. must be good.\n> - showing to them a good idea? every slide, show awesome end state thing. here's what your funding in this round gets us to. link between that and end state, logical next step. some concrete milestone that $75M gets to, clearly measurably.\n> - large amounts there.\n> - (any new info?) none of this was wildly new. basically not.\n> - (how much do you believe?) strong openai and a weak one. strong: agi is coming, has to be done in an open non-profit structure, takeoff will happen. weak: are on the cusp of progress in AI speeding up. better AI tech will be good in the ways have been good. that's good and we should fund it. no one else really funding in quite the same way and as well.\n> - fundamentally even google, which is most spend happy company, is a company at the end of the day. way we should be in this is positive externalities on the world. like science, can't monetize scientific advances that well, reasons companies are good is clear reward and incentive. in science kinda skewed. but fundamentally, not a benefit that can be captured easily. stock corporation: railroad, charge you to use it, almost all the benefits from having built the railroad, etc..\n> - progress fundamentally has to be made by non-profit, interesting direction you could go. everything i perceive with openai, race dynamics vs demis + brain + whatever, gotta get there first. another angle: they'll never do that much that's interesting.\n> - ilya: don't think it's true, that's the problem. pitch: important that not absorbed.\n> - john: ruth will get to DM like she's gotten to everyone eventually, battle between DM and ruth, ruth wins.\n> - ilya: if ruth wins then yep.\n> - john: if you want sustainable investment towards the goal. DM has most borrowed time + money. i do believe that. more correct. not the most important. think about ownership. biggest highlevel: pitch structure, more narrative oriented, significantly less detail. tactics: more interactive, clean up slides. third: get more into the mindset of the person at the ford foundation writing checks, what risks + benefits. we've been in this world long enough that are obvious to us, like why step in human efficiency.\n> - detail (??)\n\n> ok so 19 minutes ago\n> i want the fundraise to succeed 19 minutes ago\n> for the simple reason that 19 minutes ago\n> value is independent of form 19 minutes ago\n> it is better for us to become increasingly kings of this industry 18 minutes ago\n> if we give up at this juncture, i think we will never succeed 15 minutes ago\n> as before, the choice defines us 15 minutes ago\n> For sure\n> 15 minutes ago\n> also you say we should have been more ok with giving it to elon. while it's true, it's also the case he's now given it to us 15 minutes ago\n> the grand upside is here 15 minutes ago\n> i want it 15 minutes ago\n> need to stop letting distractors get tone/us 14 minutes ago\n> Being the Kings of AI is not so bad\n> 12 minutes ago\n> not so bad 12 minutes ago\n> and just remember 12 minutes ago\n> if we do it all over again 11 minutes ago\n> we know how to solve/head off all the problems we've already solved 11 minutes ago\n> we don't know how to do that with the ones we haven't 11 minutes ago\n> [Redacted]\n> also keep in mind that by allying with elon, we'll get the jim hardware 9 minutes ago\n> also don't forget there's always the \"become cerebras research\" route 9 minutes ago\n> heh 9 minutes ago\n> Jim hardware is sweet\n> 7 minutes ago\n> unique in history 6 minutes ago\n> Tesla research beats cerebras research\n> 6 minutes ago\n> (our options here) 6 minutes ago\n> or like 6 minutes ago\n> the non-profit we can build 6 minutes ago\n> if we can get the funding 6 minutes ago\n> nothing like it 6 minutes ago\n> Heh\n> 6 minutes ago\n> the us gov people have no idea what to do with their money right now 6 minutes ago\n> No doubt\n> 5 minutes ago\n> there's money sloshing around 5 minutes ago\n> Just need to find it\n> 5 minutes ago\n>\n> [Redacted]\n>\n> Every problem can be solved\n> 1 minute ago\n> yep 1 minute ago\n> value is the hard part 1 minute ago\n> let's just keep building it 1 minute ago\n> And if we're the true kings of AI\n> 1 minute ago\n> :+1::+1::+1:\n> now\n> great now\n> ok now\n> we are decided now\n> no surprises now\n> but at the very least now\n> we are now rationalized :)\n>\n> so why is this is important for people to buy into? scientific applications, we can *speed it up* and we can ensure it's not turned by google towards google purposes, owned by the world, public good. the AGI system itself for sure, but also the technologies produced along the way.\n\n## Commentary\n\nPX 163 is the **Nov 12, 2017** entry — written **six days after** Brockman told Musk in person on Nov 6 that he wanted to keep fundraising for the nonprofit. It is the cleanest version of plaintiffs' \"the founders had decided to flip to for-profit while telling Musk otherwise\" theory: in the same paragraph Brockman writes \"**we've been thinking that maybe we should just flip to a for profit. making the money for us sounds great**\" and considers Tesla, Google, or \"consortium\" money as live options. The Cerebras / \"Tesla research beats cerebras research\" / \"by allying with elon, we'll get the jim hardware\" exchange ties directly to the Brockman-Cerebras conflict-of-interest theme on [[Day 6]] (see [[Key Themes]] §\"Greg Brockman's contributions\") and to Musk's parallel Tesla AGI play. Brockman's \"**we are now rationalized :)**\" closing line is the candor plaintiffs' Mr. Molo seizes on as evidence of \"the very honest emotion that you expressed in the journal at the time\" (5/5/2026 Testimony @ ~14:08 PT). See [[Brockman Journal]] for every passage put to him on the stand and [[PX 161]] for the Nov 6, 2017 \"morally bankrupt\" entry that this one clarifies.\n\n---\n*See also:* [[Brockman Journal]] · [[Day 6|Day 6 digest]] · [[Day 7|Day 7 digest]] · [[PX 151]] · [[PX 154]] · [[PX 161]] · [[Greg Brockman]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-164", "exhibit": "PX 164", "party": "Plaintiffs", "type": "Email thread", "admitted_trial_day": "Day 6 (May 4, 2026)", "uploaded_box_pt": "2026-05-04T14:46:03", "uploader": "Someone", "pages": 4, "size_bytes": 688629, "source_pdf": "PX-164.pdf", "pdf_url": "https://media.mts-in.com/PX-164.pdf", "body_markdown": "# PX 164 — Jan 31, 2020 Brockman → Musk, \"Top AI institutions today\" / \"moral high ground\"\n\n> Greg Brockman's January 31, 2020 reply to Elon Musk (cc Sutskever, Altman, Teller, Zilis) — the \"**Our biggest tool is the moral high ground … fiduciary duty should be to humanity**\" email — written on top of Musk's same-day \"OpenAI is on a path of certain failure relative to Google\" message and an Andrej Karpathy ICLR-paper-share email.\n\n## Document type\n**Email thread, plain text, three-message chain with one embedded chart.** Brockman's Jan 31, 2018 reply [Note: header dates the thread \"Wed, 31 Jan 2018 22:56:04 -0800\" — see Anomaly note below] on top of Musk's \"Wed, Jan 31 at 2:02 PM\" message (\"OpenAI is on a path of certain failure relative to Google\") and Karpathy's \"Top AI institutions today\" forward (with an ICLR-paper-submissions plot showing Google at ~83 papers, dominating the field). Bates 2024MUSK-0004593–4596.\n\n> **Anomaly note.** The chunk metadata, the [[Brockman Journal|Brockman Journal]] wiki, and Day 6/7 testimony all label this exhibit \"**Jan 31, 2020**\" Brockman → Musk. The PDF email-header date is in fact \"**31 Jan 2018**.\" This is consistent with the **2018 fundraising-and-strategy** content (ICO discussion, three-year scaling plan, Karpathy-still-at-Tesla) — the trial witnesses appear to have been working from a transcribed citation that mis-dates the exhibit by two years. Page contents are transcribed verbatim below; readers should treat the wiki/quotes \"Jan 31, 2020\" labels as a known witness/transcript convention rather than a contradiction of the underlying document.\n\n## Logistics\n- **Trial admission:** Day 6 (May 4, 2026 — Russell direct + cross; Brockman cross by Mr. Kry). Per the Day 6 transcript table: \"PX 164 | Jan 31, 2020 Brockman → Musk: 'Our best tool is the moral high ground. To retain this, we must try our best to remain a nonprofit.'\" *(5/4/2026 Testimony @ ~11:17 PT)*. Re-read on Day 7 redirect — see [[Brockman Journal]] for the journal-sequence framing.\n- **Box upload:** 2026-05-04 14:46:03 PT — Day 6 mid-afternoon batch (clustered with PX 66, PX 153, PX 1250-16, PX 1250-52).\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~672 KB, 4 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `164.pdf`.\n\n## Transcribed text\n\n> **From:** Greg Brockman <gdb@openai.com>\n> **To:** Elon Musk <erm@spacex.com>\n> **Cc:** \"ilyasu@openai.com\" <ilyasu@openai.com>, Sam Altman <shga@ycombinator.com>, Sam Teller <steller@spacex.com>, Shivon Zilis <Shivon@spacex.com>\n> **Subject:** Re: Top AI institutions today\n> **Date:** Wed, 31 Jan 2018 22:56:04 -0800\n>\n> Hi Elon,\n>\n> Thank you for the thoughtful note. I have always been impressed by your focus on the big picture, and agree completely we must change trajectory to achieve our goals. Let's speak tomorrow, any time 4p or later will work.\n>\n> My view is that the best future will come from a major expansion of OpenAI. Our goal and mission are fundamentally correct, and that will increasingly be a superpower as AGI grows near.\n>\n> # Fundraising\n>\n> Our fundraising conversations show that:\n>\n> - Ilya and I are able to convince reputable people that AGI can really happen in the next <=10 years\n> - There's appetite for donations from those people\n> - There's *very* large appetite for investments from those people\n>\n> I respect your decision on the ICO idea, which matches the evolution of our own thinking. Sam Altman has been working on a fundraising structure that does not rely on a public offering, and we will be curious to hear your feedback.\n>\n> Of the people we've been talking to, the following people are currently my top suggestions for board members. Would also love suggestions for your top picks not on this list, and we can figure out how to approach them.\n>\n> - Reid Hoffman\n> - Gabe Newell\n> - Adam d'Angelo\n> - Jed McCaleb\n> - Herb Allen\n> - Terah Lyons (she heads Partnership on AI, originally created by Demis to steal OpenAI's thunder — would bring a lot of outside credibility)\n>\n> # The next 3 years\n>\n> Over the next 3 years, we must build 3 things:\n>\n> - Custom AI hardware (such as Jim's computer)\n> - Massive AI data center (likely multiple revs thereof)\n> - Best software team, mixing between algorithm development, public demonstrations, and safety\n>\n> We've talked the most about the custom AI hardware and AI data center. On the software front, we have a credible path (self-play in a competitive multiagent environment) which has been validated by Dota and AlphaGo. We also have identified a small but finite number of limitations in today's deep learning which are barriers to learning from human levels of experience. And we believe we uniquely are on trajectory to solving safety (at least in broad strokes) in the next three years.\n>\n> We would like to scale headcount in this way:\n>\n> - Beginning of 2017: ~40\n> - End of 2018: 100\n> - End of 2019: 300\n> - End of 2020: 900\n>\n> Note that many of DeepMind's 600+ people are not working on AGI, but instead DeepMind Health or DeepMind for Google. Every single one of our people will be working on AGI.\n>\n> ## Moral high ground\n>\n> Our biggest tool is the moral high ground. To retain this, we must:\n>\n> - **Try our best to remain a non-profit. AI is going to shake up the fabric of society, and our fiduciary duty should be to humanity.**\n> - Put increasing effort into the safety/control problem, rather than the fig leaf you've noted in other institutions. It doesn't matter who wins if everyone dies. Related to this, we need to communicate a \"**better red than dead**\" outlook — we're trying to build safe AGI, and we're not willing to destroy the world in a down-to-the-wire race to do so.\n> - Engage with government to provide trusted, unbiased policy advice — we often hear that they mistrust recommendations from companies such as Intel, Google, etc..\n> - Be perceived as a place that provides public good to the research community, and keeps the other actors honest and open via leading by example.\n>\n> # The past 2 years\n>\n> I would be curious to hear how you rate our execution over the past two years, relative to resources. In my view:\n>\n> - Over the past five years, there have two major demonstrations of working systems: AlphaZero [DeepMind] and Dota 1v1 [OpenAI]. (There are a larger number of breakthroughs of \"capabilities\" popular among practitioners, the top of which I'd say are: ProgressiveGAN [NVIDIA], unsupervised translation [Facebook], WaveNet [DeepMind], Atari/DQN [DeepMind], machine translation [Ilya at Google — now at OpenAI], generative adversarial network [Ian Goodfellow at grad school — now at Google], variational autoencoder [Durk at grad school — now at OpenAI], AlexNet [Ilya at grad school — now at OpenAI].) We benchmark well on this axis.\n> - We grew very rapidly in 2016, and in 2017 iterated to a working management structure. We're now ready to scale massively, given the resources. We lose people on comp currently, but pretty much *only* on comp. I've been resuming the style of recruiting I did in the early days, and believe I can exceed those results.\n> - We have the most talent dense team in the field, and we have the reputation for it as well.\n> - We don't encourage paper writing, and so paper acceptance isn't a measure we optimize. For the ICLR chart Andrej sent, I'd expect our (accepted papers)/(people submitting papers) to be the highest in the field.\n>\n> - gdb\n\n> **From:** Elon Musk <erm@spacex.com>\n> **To:** Greg Brockman <gdb@openai.com>, Ilya Sutskever <ilyasu@openai.com>, Sam Altman <shga@ycombinator.com>\n> **Cc:** Sam Teller <steller@spacex.com>, Shivon Zilis <Shivon@spacex.com>\n> **Subject:** Fwd: Top AI institutions today\n> **Date:** Wed, Jan 31, 2018 at 2:02 PM\n>\n> OpenAI is on a path of certain failure relative to Google. There obviously needs to be immediate and dramatic action or everyone except for Google will be consigned to irrelevance.\n>\n> I have considered the ICO approach and will not support it. In my opinion, that would simply result in a massive loss of credibility for OpenAI and everyone associated with the ICO. If something seems too good to be true, it is. This was, in my opinion, an unwise diversion.\n>\n> The only paths I can think of are a major expansion of OpenAI and a major expansion of Tesla AI. Perhaps both simultaneously. The former would require a major increase in funds donated and highly credible people joining our board. The current board situation is very weak.\n>\n> I will set up a time for us to talk tomorrow. To be clear, I have a lot of respect for your abilities and accomplishments, but I am not happy with how things have been managed. That is why I have had trouble engaging with OpenAI in recent months. Either we fix things and my engagement increases a lot or we don't and I will drop to near zero and publicly reduce my association. I will not be in a situation where the perception of my influence and time doesn't match the reality.\n>\n> Begin forwarded message:\n\n> **From:** Andrej Karpathy <akarpathy@tesla.com>\n> **To:** Elon Musk <erm@tesla.com>\n> **Cc:** Shivon Zilis <shivon@tesla.com>\n> **Subject:** Top AI institutions today\n> **Date:** January 31, 2018 at 1:20:42 PM PST\n>\n> The ICLR conference (which is the top deep learning – specific conference (NIPS is larger, but more diffuse)) released their decisions for accepted/rejected papers, and someone made some nice plots that show where the current deep learning / AI research happens at. It's an imperfect measure because not every company might prioritize paper publications, but it's indicative.\n> >\n> > Here's a plot that shows the total number of papers (broken down by oral/poster/workshop/rejected) from any institution:\n> >\n> > [Embedded image: bar chart of ICLR submissions by institution. Google leads at ~83. Berkeley/Stanford ~28. CMU ~32. Microsoft, Facebook, MIT, ETH, IBM, Toronto Univ., Oxford, NYU, USC, etc., follow. OpenAI is in the middle of the pack at single digits.]\n> >\n> > Long story short, Google is dominating with 83 paper submissions. The academic institutions (Berkeley / Stanford / CMU / MIT) are next, in 20-30 ranges each.\n> >\n> > Just thought it was an interesting snapshot of where all the action is today. The full data is here: http://webia.lip6.fr/~pajot/dataviz.html\n> >\n> > -Andrej\n\n## Commentary\n\nPX 164 is the email plaintiffs put on the screen as the [[Brockman Journal|\"moral high ground\"]] receipt — \"**Our biggest tool is the moral high ground. To retain this, we must … try our best to remain a non-profit. AI is going to shake up the fabric of society, and our fiduciary duty should be to humanity.**\" Mr. Molo on Day 7 read this same passage in sequence with the Brockman journal entries (PX 151, 154, 161, 163; DX 1252) as the **2018 echo of the founders' 2017 nonprofit-mission framing** — ammunition for plaintiffs' \"[[Key Themes|stole a charity]]\" theme. Defense's read is that the surrounding context cuts the other way: Brockman is responding to a Musk message (preserved on this same exhibit) declaring OpenAI on \"**a path of certain failure relative to Google**\" and threatening to \"drop to near zero and publicly reduce my association\" — i.e., the \"moral high ground\" passage is part of a fundraising-and-strategy email written *to keep Musk engaged* at a moment when Musk was leveraging engagement, not a charter-of-conscience entry. This is also the email in which Brockman teases the **2017 ICO discussion** Musk had vetoed and the Sam-Altman-led **alternative fundraising structure that does not rely on a public offering** — i.e., the architectural seed of the 2019 capped-profit LP ([[DX-827|DX 827]]). Note also the 900-headcount-by-2020 plan (\"Every single one of our people will be working on AGI\") and the embedded Karpathy ICLR plot — the same Karpathy who, a few months later, would be the subject of [[DX-619|DX 619]]'s poaching-coordination texts. See [[Brockman Journal]] for the cross-walk and the **note that this exhibit is an email, not a journal entry**.\n\n---\n*See also:* [[Day 6|Day 6 digest]] · [[Day 7|Day 7 digest]] · [[Brockman Journal]] · [[Greg Brockman]] · [[Elon Musk]] · [[PX 151]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-167", "exhibit": "PX 167", "party": "Plaintiffs", "type": "Text messages", "admitted_trial_day": "Day 8 (May 6, 2026)", "uploaded_box_pt": "2026-05-06T14:07:56", "uploader": "Someone", "pages": 3, "size_bytes": 313219, "source_pdf": "PX-167.pdf", "pdf_url": "https://media.mts-in.com/PX-167.pdf", "body_markdown": "# PX 167 — Feb 17, 2018 Altman–Zilis text messages re Musk poaching OpenAI staff for Tesla\n\n> Forensic Cellebrite \"Short Message Report\" of 17 iMessage messages between Sam Altman and Shivon Zilis on February 17, 2018, in which Altman tells Zilis that Musk \"called people to try to recruit them\" away from OpenAI to Tesla.\n\n## Document type\n**Text messages (forensic extract).** Cellebrite-style \"Short Message Report\" header with conversation hash `97cf579ac2f61c9488dc8cce96ab8721`, 17 messages, 2 participants (Sam Altman; Shivon Zilis), date range 2/17/2018. Apple iMessage. Phone numbers redacted. Bates ZILIS-0002197 through ZILIS-0002199.\n\n## Logistics\n- **Trial admission:** Day 8 (May 6, 2026) — Murati live; Zilis live; Toner video begins.\n- **Box upload:** 2026-05-06 14:07:56 PT.\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~306 KB, 3 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `167.pdf`.\n\n## Transcribed text\n\n> **Short Message Report**\n> Conversations: 1 / Participants: 2 / Total Messages: 17 / Date Range: 2/17/2018\n>\n> **Outline of Conversations:** `97cf579ac2f61c9488dc8cce96ab8721` — 17 messages on 2/17/2018 — Altman (+[redacted]) · Shivon Zilis (+[redacted])\n\n**Messages in chronological order (times shown in GMT +00:00):**\n\n> **SZ — Shivon Zilis** — 2/17/2018, 12:27 AM\n> When do you think you'll get to a draft and should I coordinate with Jack or all good?\n>\n> **A — Altman** — 2/17/2018, 5:48 PM\n> btw, w/number of people elon has reached out to, basically most of openai knows at this point...\n>\n> **SZ — Shivon Zilis** — 2/17/2018, 6:06 PM\n> Neuralink or OpenAI people?\n>\n> **A — Altman** — 2/17/2018, 6:06 PM\n> openai\n>\n> **A — Altman** — 2/17/2018, 6:06 PM\n> calling people to try to recruit them\n>\n> **A — Altman** — 2/17/2018, 6:06 PM\n> who are then asking question to other people if they got similar calls\n>\n> **SZ — Shivon Zilis** — 2/17/2018, 6:06 PM\n> I didn't know he reached out to OpenAI people?\n>\n> **SZ — Shivon Zilis** — 2/17/2018, 6:07 PM\n> Minus Woj and I presumed Scott\n>\n> **SZ — Shivon Zilis** — 2/17/2018, 6:07 PM\n> Hm ok\n>\n> **A — Altman** — 2/17/2018, 6:07 PM\n> he told scott he when he reached out to him that he didnt think openai could succeed and that he was talking to the 4-5 best people to get them to tesla\n>\n> **A — Altman** — 2/17/2018, 6:08 PM\n> and i think he reached out to some of the dota team? but i only heard part of the conversation this morning\n>\n> **SZ — Shivon Zilis** — 2/17/2018, 6:11 PM\n> Have no idea on DoTA tbh. Haven't been with him\n>\n> **SZ — Shivon Zilis** — 2/17/2018, 6:23 PM\n> This interaction didn't feel right. Not urgent, but let's chat briefly next week about some general stuff. I would love to keep helping you as much as I can but important to figure out the correct frameworks of trust for each other\n>\n> **A — Altman** — 2/17/2018, 6:24 PM\n> ok--sincere apologies, certainly didnt mean anything that felt badly, but on a re-read i totally get what happened. my goal was simply to give you a heads up on a chance that news leaks out sooner, given i know that there's a lot of sensitivity there\n>\n> **A — Altman** — 2/17/2018, 6:24 PM\n> im around any time this weekend if you'd like to chat, or next week.\n>\n> **SZ — Shivon Zilis** — 2/17/2018, 6:29 PM\n> Don't sweat it too much. Text is super lossy! We'll just chat on it in person next week\n>\n> **A — Altman** — 2/17/2018, 6:31 PM\n> kk\n\n## Commentary\n\nPX 167 is plaintiffs' contemporaneous corroboration of the Tesla-poaching theme that became one of the [[Key Themes|\"Tesla / xAI as Musk's own AGI play\"]] central pillars on cross. Altman's near-real-time message — \"**he told scott he when he reached out to him that he didnt think openai could succeed and that he was talking to the 4-5 best people to get them to tesla**\" — runs parallel to the [[DX-761|Musk-to-Zilis Feb 2018 email]] cited in [[Key Themes]] (\"we are going to actively try to move three or four people from OpenAI to Tesla\") and to defense's poach-list testimony on [[Day 8|Day 8]] from Karpathy. The exhibit is doubly useful to plaintiffs because it shows Zilis — Musk's information conduit into OpenAI — *learning of the poaching from Altman, not Musk*, and reacting that \"this interaction didn't feel right.\" That detail cuts against the defense's \"Zilis was Musk's eyes and ears\" framing and supports plaintiffs' theory that Musk treated OpenAI staff as Tesla recruitment pipeline while still serving on the OpenAI board. See also [[Day 8|Day 8 digest]] for the live cross of Zilis and the [[DX-758|\"OpenAI possible scenarios\"]] document.\n\n---\n*See also:* [[Day 8|Day 8 digest]] · [[Shivon Zilis]] · [[Sam Altman]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-17", "exhibit": "PX 17", "party": "Plaintiffs", "type": "Blog post", "admitted_trial_day": "Day 4 (April 30, 2026)", "uploaded_box_pt": "2026-04-30T15:26:48", "uploader": "Someone", "pages": 4, "size_bytes": 3592821, "source_pdf": "PX-17.pdf", "bates": "2024MUSK-0014466–0014469", "pdf_url": "https://media.mts-in.com/PX-17.pdf", "body_markdown": "# PX 17 — \"Introducing OpenAI\" blog post (December 11, 2015)\n\n> The original December 11, 2015 OpenAI launch announcement on openai.com — the public-facing founding declaration that OpenAI was a non-profit \"unconstrained by a need to generate financial return\" backed by ~$1 billion in committed funding.\n\n## Document type\n**Blog post.** Public-facing OpenAI.com launch announcement, dated December 11, 2015, filed under \"Company.\" Authors: Greg Brockman, Ilya Sutskever, OpenAI. Four pages, including images and a \"Culture & Careers / Community / 2015\" footer.\n\n## Logistics\n- **Trial admission:** Day 4 (April 30, 2026) — used during Musk cross + redirect; Birchall direct + cross. The \"OPEN means open source\" / \"non-profit unconstrained by financial return\" textual anchor that runs through both Musk's direct (Day 2) and the cross (Day 3–4).\n- **Box upload:** 2026-04-30 15:26:48 PT — uploaded the same afternoon as PX 86, PX 100, and other Day 4 cross exhibits.\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~3.6 MB, 4 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `017.pdf`.\n\n## Transcribed text\n\n> **December 11, 2015 — Company**\n>\n> # Introducing OpenAI\n>\n> [Listen to article — 3:31 — Share]\n>\n> OpenAI is a non-profit artificial intelligence research company. Our goal is to advance digital intelligence in the way that is most likely to benefit humanity as a whole, unconstrained by a need to generate financial return. Since our research is free from financial obligations, we can better focus on a positive human impact.\n>\n> We believe AI should be an extension of individual human wills and, in the spirit of liberty, as broadly and evenly distributed as possible. The outcome of this venture is uncertain and the work is difficult, but we believe the goal and the structure are right. We hope this is what matters most to the best in the field.\n>\n> ## Background\n>\n> Artificial intelligence has always been a surprising field. In the early days, people thought that solving certain tasks (such as chess) would lead us to discover human-level intelligence algorithms. However, the solution to each task turned out to be much less general than people were hoping (such as doing a search over a huge number of moves).\n>\n> The past few years have held another flavor of surprise. An AI technique explored for decades, deep learning, started achieving state-of-the-art results in a wide variety of problem domains. In deep learning, rather than hand-code a new algorithm for each problem, you design architectures that can twist themselves into a wide range of algorithms based on the data you feed them.\n>\n> This approach has yielded outstanding results on pattern recognition problems, such as recognizing objects in images, machine translation, and speech recognition. But we've also started to see what it might be like for computers to be creative, to dream, and to experience the world.\n>\n> ## Looking forward\n>\n> AI systems today have impressive but narrow capabilities. It seems that we'll keep whittling away at their constraints, and in the extreme case they will reach human performance on virtually every intellectual task. It's hard to fathom how much human-level AI could benefit society, and it's equally hard to imagine how much it could damage society if built or used incorrectly.\n>\n> ## OpenAI\n>\n> Because of AI's surprising history, it's hard to predict when human-level AI might come within reach. When it does, it'll be important to have a leading research institution which can prioritize a good outcome for all over its own self-interest.\n>\n> We're hoping to grow OpenAI into such an institution. As a non-profit, our aim is to build value for everyone rather than shareholders. Researchers will be strongly encouraged to publish their work, whether as papers, blog posts, or code, and our patents (if any) will be shared with the world. We'll freely collaborate with others across many institutions and expect to work with companies to research and deploy new technologies.\n>\n> OpenAI's research director is Ilya Sutskever, one of the world experts in machine learning. Our CTO is Greg Brockman, formerly the CTO of Stripe. The group's other founding members are world-class research engineers and scientists: Trevor Blackwell, Vicki Cheung, Andrej Karpathy, Durk Kingma, John Schulman, Pamela Vagata, and Wojciech Zaremba. Pieter Abbeel, Yoshua Bengio, Alan Kay, Sergey Levine, and Vishal Sikka are advisors to the group. OpenAI's co-chairs are Sam Altman and Elon Musk.\n>\n> Sam, Greg, Elon, Reid Hoffman, Jessica Livingston, Peter Thiel, Amazon Web Services (AWS), Infosys, and YC Research are donating to support OpenAI. In total, these funders have committed $1 billion, although we expect to only spend a tiny fraction of this in the next few years.\n>\n> You can follow us on Twitter at @OpenAI.\n>\n> ---\n> Culture & Careers · Community · 2015\n>\n> Authors: Greg Brockman, Ilya Sutskever, OpenAI\n\n## Commentary\n\nPX 17 is the textual cornerstone of plaintiffs' \"stole a charity\" theory: the public, world-facing December 11, 2015 declaration that OpenAI was \"a non-profit artificial intelligence research company\" whose research would be \"unconstrained by a need to generate financial return,\" whose patents \"will be shared with the world,\" and whose mission was \"to build value for everyone rather than shareholders.\" The \"$1 billion\" line — \"these funders have committed $1 billion\" — is the source of the defense's most-used cross point that Musk eventually contributed only $38 million ([[Key Themes]] — \"$1B / $38M gap\"), but plaintiffs use the same line as the original public solicitation that anchors the charitable-trust claim. The post is also the document the judge had in mind when she pointedly told OpenAI counsel after openings that \"you do not take inconsistent positions in front of the[ PTO]\" about whether \"OPEN\" meant \"open source\" — patents-shared-with-the-world and code-published-as-blog-posts is the December 2015 promise plaintiffs say was abandoned. See [[Day 2|Musk's direct]] and [[Day 4|Day 4 digest]] for the cross-examination treatment.\n\n---\n*See also:* [[Day 4|Day 4 digest]] · [[PX 70]] · [[PX 156]] · [[PX 157]] · [[Greg Brockman]] · [[Ilya Sutskever]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-200", "exhibit": "PX 200", "party": "Plaintiffs", "type": "Contract", "admitted_trial_day": "Day 7 (May 5, 2026)", "uploaded_box_pt": "2026-05-05T16:29:55", "uploader": "Someone", "pages": 68, "size_bytes": 12394121, "source_pdf": "PX-200.pdf", "bates": "OPENAI_MUSK00013411–00013478", "pdf_url": "https://media.mts-in.com/PX-200.pdf", "body_markdown": "# PX 200 — OpenAI, L.P. Limited Partnership Agreement (October 10, 2018)\n\n> The 68-page founding LPA of OpenAI, L.P., the for-profit Delaware limited partnership Musk's lawsuit calls the looted nonprofit's vehicle of capture — the document that contains the famous \"purple box,\" the \"Core Mission\" subordination clause, the Target Redemption Amount waterfall, and the original 2018 Charter as Exhibit A.\n\n## Document type\n**Contract — Limited Partnership Agreement.** A 68-page Delaware LPA dated October 10, 2018, executed by David Lansky as Manager of OpenAI GP, L.L.C. (General Partner) and as Attorney-in-Fact for the Limited Partners. Counterparty/drafter: Goodwin Procter LLP (Section 10.18). The OpenAI, Inc. Charter (\"primary fiduciary duty is to humanity\") is incorporated as Exhibit A. The \"Lead Investor\" is identified by name as the Aphorism Foundation (Section 1.1, \"Lead Investor\"). Bates OPENAI_MUSK00013411–13478. Marked \"Highly Confidential.\"\n\n## Logistics\n- **Trial admission:** Day 7 (May 5, 2026) — Brockman defense direct + Mr. Molo recross; Robert Wu 30(b)(6) deposition. The structural foundation for everything Wu testified about: the LP waterfall, the Target Redemption Amounts, Microsoft's 2019 and 2023 redemption stacks, the \"$250 billion\" total target redemption back-of-the-napkin figure, and the AGI carveout. See [[Key Themes]] §\"Day 7 — the Microsoft economics.\"\n- **Box upload:** 2026-05-05 16:29:55 PT — Day 7 late-afternoon batch (within ten minutes of [[PX 212]]).\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~12.4 MB, 68 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `200.pdf`.\n\n## Transcribed text\n\n> **LIMITED PARTNERSHIP AGREEMENT OF OPENAI, L.P. — A DELAWARE LIMITED PARTNERSHIP — October 10, 2018**\n\n### The \"purple box\" (front matter, Page 1)\n\n> **IMPORTANT**\n>\n> **\\*\\*Investing in OpenAI, L.P. is a *high-risk investment*\\*\\***\n> **\\*\\*Investors could lose their capital contribution and not see any return\\*\\***\n> **\\*\\*It would be wise to view any investment in OpenAI, L.P. in the spirit of a donation, with the understanding that it may be difficult to know what role money will play in a post-AGI world\\*\\***\n>\n> The Partnership exists to advance OpenAI, Inc.'s mission of ensuring that safe artificial general intelligence is developed and benefits all of humanity. The General Partner's duty to this mission and the principles advanced in the OpenAI, Inc. Charter take precedence over any obligation to generate a profit. The Partnership may never make a profit, and the General Partner is under no obligation to do so. The General Partner is free to re-invest any or all of the Partnership's cash flow into research and development activities and/or related expenses without any obligation to the Limited Partners. See Section 6.4 for additional details.\n\n### Section 1.1 — Key Definitions (selected)\n\n> **AGI** shall have the meaning set forth in Section 2.3(a). [defined as \"a safe artificial general intelligence — meaning a highly autonomous system that outperforms humans at most economically valuable work\"]\n>\n> **Commencement of the Redemption Period** shall mean the earlier of: (x) the time at which both (i) the General Partner has made the Declaration of Sufficient AGI, and (ii) the General Partner has determined, in its Sole Discretion, that distributions will commence pursuant to Section 5.1(b)(ii); or (y) the time at which the General Partner and a Majority-In-Interest of the Limited Partners elect (each acting in their Sole Discretion) to commence distributions pursuant to Section 5.1(b)(ii).\n>\n> **Core Mission** shall have the meaning set forth in Section 2.3(a).\n>\n> **Declaration of Sufficient AGI** shall mean the effective time of the General Partner's determination, in its reasonable discretion, that (x) AGI has been created that has the capability to generate the Target Redemption Amount for each Partner; and (y) it has the ability and authority to direct the AGI to generate such Target Redemption Amounts.\n>\n> **Employee Vehicle** shall mean OpenAI Holdings, L.P., a Delaware limited partnership.\n>\n> **General Partner** shall mean OpenAI GP, L.L.C., a Delaware limited liability company.\n>\n> **Lead Investor** shall mean the Aphorism Foundation.\n>\n> **Non-Profit** shall mean OpenAI, Inc., a Delaware nonprofit nonstock corporation.\n>\n> **Non-Profit Charter** shall mean that certain \"charter\" of the Non-Profit (a copy of which, as of the Initial Closing, is attached as Exhibit A) as such charter may be supplemented, modified, changed or replaced by the Non-Profit from time to time in its Sole Discretion.\n>\n> **Target Redemption Amount** shall mean, for each Partner, the dollar amount specified for such Member under the heading \"Target Redemption Amount\" on Schedule A, and increased as follows: (x) for the partial Fiscal Year that begins on the Initial Closing, and the first ten years occurring after the Initial Closing, the Target Redemption Amount shall be increased by the rate of inflation for such full or partial Fiscal Year… (the \"Inflation Rate\"); and (y) for each full or partial Fiscal Year that begins after the tenth anniversary of the Initial Closing, the Target Redemption Amount for each Partner shall be increased by the greater of the Inflation Rate for such full or partial Fiscal Year, and a percent equal to LIBOR plus 2 percent.\n\n### Section 2.3 — Purpose and Scope\n\n> Within the meaning and for purposes of the Act, the purpose and scope of the Partnership shall be to:\n>\n> (a) Use the principles expounded in the Non-Profit Charter to advance the Non-Profit's mission of pursuing (i) the creation, development and refinement of a safe artificial general intelligence — meaning a highly autonomous system that outperforms humans at most economically valuable work — (\"AGI\"); and (ii) the dissemination and distribution of the benefits of such AGI among all of humanity (collectively, the \"Core Mission\").\n>\n> (b) Subordinate to the Core Mission, seek income and gain through selective licensing, commercialization or other monetization of technologies, scientific breakthroughs or other similar discoveries that are achieved in the pursuit, or furtherance, of the Core Mission;\n>\n> (c) Temporarily invest cash in Idle Funds Investments; and\n>\n> (d) Engage in any other lawful activities (including activities described in Section 6.2) determined by the General Partner to be necessary or advisable in furtherance of the foregoing activities.\n\n### Section 6.4 — Subordination of Partnership and Limited Partner Interests to Core Mission; Modification of Fiduciary Duties (excerpt — original ALL CAPS preserved)\n\n> NOTWITHSTANDING ANY PROVISION OF THIS AGREEMENT TO THE CONTRARY, THE PARTNERS ACKNOWLEDGE AND AGREE THAT, AS DETERMINED BY THE GENERAL PARTNER IN ITS SOLE DISCRETION, THE INTERESTS OF THE PARTNERSHIP AND THE LIMITED PARTNERS SHALL BE SUBORDINATE TO THE CORE MISSION. UNDER NO CIRCUMSTANCES SHALL THE GENERAL PARTNER BE LIABLE TO THE PARTNERSHIP OR THE LIMITED PARTNERS IN CONSEQUENCE OF HAVING TAKEN ANY ACTION INTENDED TO FURTHER THE CORE MISSION, EVEN IF SUCH ACTION CONFLICTS WITH THE INTERESTS OF OR, ACTUALLY RESULTS IN ADVERSE CONSEQUENCES TO, THE PARTNERSHIP OR THE LIMITED PARTNERS. … MOREOVER, THE DUTIES AND OBLIGATIONS OF THE GENERAL PARTNER ARISING UNDER THIS AGREEMENT OR OTHERWISE IN ITS CAPACITY AS SUCH SHALL BE SUBORDINATE TO THE EFFORTS OF THE GP RELATED PERSONS TO FULFILL THE CORE MISSION. … NONE OF THE GENERAL PARTNER, ANY GP RELATED PERSON … OR ANY AFFILIATE OR ASSOCIATE OF ANY OF THE FOREGOING, SHALL BE DEEMED TO HAVE VIOLATED ANY FIDUCIARY OR OTHER DUTY TO ANY PERSON IN CONSEQUENCE OF HAVING PLACED THE CORE MISSION AHEAD OF THE INTERESTS OF THE PARTNERSHIP OR THE LIMITED PARTNERS.\n>\n> Without limitation on the generality of the forgoing… the General Partner may cause the Partnership to take any of the following actions that the General Partner, reasonably and in good faith, determines could support or further the Core Mission:\n>\n> (i) Spend, invest or otherwise use any of the Partnership's cash or property in any manner, even if such use will delay or impair the economic returns realized by the Limited Partners…\n>\n> (xi) Retain any and all cash or property held by the Partnership as the General Partner deems necessary to fund the acquisition of any business (including one or more large publicly-traded businesses) that the General Partner reasonably and in good faith determines could achieve AGI (prior to the Declaration of Sufficient AGI), even if the manner of such achievement of AGI conflicts with the Core Mission.\n>\n> If the General Partner declares that any action or inaction, pursuant to the foregoing or otherwise, is in furtherance of or supports the Core Mission, such action or inaction shall automatically be deemed to be undertaken reasonably and in good faith in support of the Core Mission, and within the scope of the General Partner's authority under this Agreement, unless there is material clear and convincing evidence that the General Partner had actual knowledge that such action or inaction affirmatively had no realistic possibility of supporting or furthering the Core Mission.\n\n### Section 5.1(b)(ii) — Distribution Waterfall, From and After Commencement of the Redemption Period\n\n> (A) First, to the First Close Limited Partners (including the Non-Profit with respect to its Limited Partner Investment Interest) until each such Partner has received aggregate distributions pursuant to this Section 5.1(b)(ii)(A) equal to its Capital Contributions.\n>\n> (B) Next, any remaining items of cash or property comprising such distribution shall be apportioned and distributed (1) 25 percent, in proportion to each Partner's Target Redemption Amount, to the First Close Limited Partners (including the Non-Profit with respect to its Limited Partner Investment Interest), and to the Employee Vehicle; and (2) 75 percent, in proportion to their respective amounts remaining to be distributed pursuant to this Section 5.1(b)(ii)(B), to the Second Close Limited Partners, until each such Second Close Limited Partner has received aggregate distributions pursuant to this Section 5.1(b)(ii)(B) at least equal to its Capital Contributions.\n>\n> (C) Next, to the First Close Limited Partners (including the Non-Profit with respect to its Limited Partner Investment Interest), and to the Employee Vehicle, and to the Second Close Limited Partners, until the time that each has received aggregate distributions pursuant to this Section 5.1(b)(ii)(C) equal to its respective Target Redemption Amount…\n>\n> (D) Next, all remaining items of cash and property shall be distributed to the Non-Profit in its capacity as a Limited Partner.\n\n### Schedule A — Partner Information (Initial)\n\n> **GENERAL PARTNER:** OpenAI GP, L.L.C., a Delaware limited liability company. 3180 18th Street, San Francisco, CA 94110. Attn: Mr. David Lansky, General Counsel. Telephone: 800-217-3145. E-mail: david@openai.com\n>\n> **LIMITED PARTNERS:**\n> - **OpenAI, Inc.,** a Delaware nonprofit nonstock corporation. Capital Commitment: To be determined.* Target Redemption Amount: [blank]\n> - **OpenAI Holdings, L.P.,** a Delaware limited partnership (the Employee Vehicle). Capital Commitment: $0. Target Redemption Amount: **$100 Billion**.\n>\n> *\\* Fair market value to be determined by the General Partner and inserted following Initial Closing.*\n\n### Exhibit A — OpenAI, Inc. Charter (attached)\n\n> **OpenAI Charter** — OpenAI's mission is to ensure that artificial general intelligence (AGI) — by which we mean highly autonomous systems that outperform humans at most economically valuable work — benefits all of humanity. We will attempt to directly build safe and beneficial AGI, but will also consider our mission fulfilled if our work aids others to achieve this outcome. To that end, we commit to the following principles:\n>\n> **Broadly Distributed Benefits.** We commit to use any influence we obtain over AGI's deployment to ensure it is used for the benefit of all, and to avoid enabling uses of AI or AGI that harm humanity or unduly concentrate power.\n>\n> **Our primary fiduciary duty is to humanity.** We anticipate needing to marshal substantial resources to fulfill our mission, but will always diligently act to minimize conflicts of interest among our employees and stakeholders that could compromise broad benefit.\n>\n> **Long-Term Safety.** We are committed to doing the research required to make AGI safe, and to driving the broad adoption of such research across the AI community. We are concerned about late-stage AGI development becoming a competitive race without time for adequate safety precautions. Therefore, if a value-aligned, safety-conscious project comes close to building AGI before we do, we commit to stop competing with and start assisting this project. We will work out specifics in case-by-case agreements, but a typical triggering condition might be \"a better-than-even chance of success in the next two years.\"\n>\n> **Technical Leadership.** To be effective at addressing AGI's impact on society, OpenAI must be on the cutting edge of AI capabilities — policy and safety advocacy alone would be insufficient. We believe that AI will have broad societal impact before AGI, and we'll strive to lead in those areas that are directly aligned with our mission and expertise.\n>\n> **Cooperative Orientation.** We will actively cooperate with other research and policy institutions; we seek to create a global community working together to address AGI's global challenges. We are committed to providing public goods that help society navigate the path to AGI. Today this includes publishing most of our AI research, but we expect that safety and security concerns will reduce our traditional publishing in the future, while increasing the importance of sharing safety, policy, and standards research.\n\n### [Sections 1 (remaining definitions), 2.4–2.7, 3 (Capitalization beyond 3.1), 4 (Profits and Losses), 5.1(a) (Tax Distributions), 5.2–5.4, 6 (Administration apart from 6.4), 7 (Transfers and Withdrawals), 8 (Dissolution and Liquidation), 9 (Liability and Indemnification), and 10 (General Provisions including 10.13 dispute-resolution arbitration in Santa Clara County) — the remainder of the agreement is standard Delaware-LP boilerplate plus customary OpenAI carveouts (Holding Vehicles, Feeder Vehicles, partnership tax representative, FATCA/AML compliance, Whistleblower protections, Goodwin Procter conflict waiver). Pages 5–58 of the body text and signature/Schedule A pages 64–65 are not transcribed verbatim above.]\n\n### Signature\n\n> IN WITNESS WHEREOF, the parties have executed this Limited Partnership Agreement of OpenAI, L.P., a Delaware limited partnership, as of the date first above written.\n>\n> **GENERAL PARTNER:** OpenAI GP, L.L.C., a Delaware limited liability company. By: *(signature)* Name: David Lansky. Title: Manager.\n>\n> **LIMITED PARTNERS:** THE LIMITED PARTNERS LISTED ON SCHEDULE A HERETO. By: OpenAI GP, L.L.C., a Delaware limited liability company, Title: Attorney-in-Fact. By: *(signature)* Name: David Lansky. Title: Manager.\n\n## Commentary\n\nPX 200 is the founding contract of the OpenAI for-profit. Plaintiffs treat it as the first piece of paper that operationalized the breach of charitable trust: a Delaware limited partnership in which the General Partner is contractually empowered (Section 6.4) to subordinate Partnership and Limited Partner interests to the \"Core Mission\" — but only at the General Partner's \"Sole Discretion,\" with fiduciary duties to LPs explicitly waived where Core Mission is invoked. The \"purple box\" cousin lives at front-matter page 1 here (\"It would be wise to view any investment in OpenAI, L.P. in the spirit of a donation\"), which the defense's [[Key Themes]] §\"The 2018 'purple box' term sheet\" reading uses as evidence that no investor (including any Musk vehicle, had Musk invested) could have been deceived. Plaintiffs' read is the inverse: the **mechanism** — Target Redemption Amounts that grow at LIBOR+2% after year ten, a residual to the nonprofit only **after** the LP stack and the Employee Vehicle's $100B redemption-amount pole are paid out — is the \"looting\" architecture itself. The OpenAI Charter attached as Exhibit A is the document Brockman re-drafted as [\"Charter 2.0\" in December 2023 (PX 359)](#) to remove \"Our primary fiduciary duty is to humanity\" — see [[Key Themes]] §\"Charter 2.0.\" Wu's Day 7 testimony (the $250B back-of-the-napkin redemption stack, the 6× Microsoft return, the 20% annual escalator, the AGI carveout dissolving in the Watershed MOU) is the empirical reading of the waterfall this contract installs.\n\n---\n*See also:* [[Day 7|Day 7 digest]] · [[PX 212]] · [[PX 359]] · [[Robert Wu]] · [[Greg Brockman]] · [[Brockman Journal]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-201", "exhibit": "PX 201", "party": "Plaintiffs", "type": "Corporate governance document — Amended and Restated Limited Partnership Agreement (with attached Charter and Escrow Agreement)", "admitted_trial_day": "Day 7 (May 5, 2026)", "uploaded_box_pt": "2026-05-05T17:55:03", "uploader": "Someone", "pages": 93, "size_bytes": 29815002, "source_pdf": "PX-201.pdf", "split_parts": "0201.part1of3.pdf, 0201.part2of3.pdf, 0201.part3of3.pdf", "bates": "MSFT_MUSK000059951–MSFT_MUSK000060043 (approx.)", "pdf_url": "https://media.mts-in.com/PX-201.pdf", "body_markdown": "# PX 201 — OpenAI, L.P. Amended and Restated Limited Partnership Agreement (July 2, 2019)\n\n> The 2019 OpenAI, L.P. LPA — the document that converted the OpenAI Capped-Profit structure into Microsoft's vehicle, with Microsoft executing as the \"Converted Second Close Limited Partner,\" signed by Satya Nadella for Microsoft and David Lansky as Manager of OpenAI GP, L.L.C. Includes the OpenAI Inc. Charter (Exhibit A) and the SunTrust escrow/trust deposit agreement (Exhibit B).\n\n## Document type\n**Corporate governance document — multi-doc compilation.** Bates-stamped Microsoft production (MSFT_MUSK series). The package contains: (i) the **Amended and Restated Limited Partnership Agreement of OpenAI, L.P.** dated July 2, 2019 (Sections 1–10, ~66 numbered pages); (ii) the executed signature page (Microsoft + OpenAI GP); (iii) **Schedule A — Partner Information** listing Capital Commitments and Target Redemption Amounts for each LP; (iv) **Exhibit A — the OpenAI Charter** (April 9, 2018, printed from openai.com/charter); and (v) **Exhibit B — the Escrow and Trust Deposit Agreement** of July 2, 2019 with SunTrust Bank as Custodian.\n\n## Logistics\n- **Trial admission:** Day 7 (May 5, 2026). Used during the Robert Wu 30(b)(6) deposition designations and the Brockman recross — the LPA is the structural document underneath plaintiffs' \"$250B target-redemption stack\" / Microsoft-economics theory developed on Day 7.\n- **Box upload:** 2026-05-05 17:55:03 PT — Day 7 evening batch (consistent with end-of-day evidentiary upload).\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~28.4 MB, 93 pages. **Note:** original PDF exceeded the 20 MB Read tool cap and was split into three parts in `Elon Musk, et al. v. Samuel Altman, et al - split/`.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `0201.pdf`.\n\n## Transcribed text\n\n> **AMENDED AND RESTATED**\n> **LIMITED PARTNERSHIP AGREEMENT OF OPENAI, L.P.**\n> **A DELAWARE LIMITED PARTNERSHIP**\n> **July 2, 2019**\n>\n> **IMPORTANT**\n>\n> *Investing in OpenAI, L.P. is a high-risk investment*\n> *Investors could lose their capital contribution and not see any return*\n> *It would be wise to view any investment in OpenAI, L.P. in the spirit of a donation, with the understanding that it may be difficult to know what role money will play in a post-AGI world*\n>\n> The Partnership exists to advance OpenAI, Inc.'s mission of ensuring that safe artificial general intelligence is developed and benefits all of humanity. The General Partner's duty to this mission and the principles advanced in the OpenAI, Inc. Charter take precedence over any obligation to generate a profit. The Partnership may never make a profit, and the General Partner is under no obligation to do so. The General Partner is free to re-invest any or all of the Partnership's cash flow into research and development activities and/or related expenses without any obligation to the Limited Partners. See Section 6.4 for additional details.\n>\n> *[HIGHLY CONFIDENTIAL — MSFT_MUSK000059951]*\n\n> **SECTION 1 — DEFINITIONS (selected):**\n>\n> - **Converted Second Close Limited Partner** means **Microsoft Corporation**.\n> - **Core Mission** (per § 2.3(a)): \"Use the principles expounded in the Non-Profit Charter to advance the Non-Profit's mission of pursuing (i) the creation, development and refinement of a safe artificial general intelligence — meaning a highly autonomous system that outperforms humans at most economically valuable work — ('AGI'); and (ii) the dissemination and distribution of the benefits of such AGI among all of humanity.\"\n> - **Major Decisions** (the LP-consent gating list) include: \"(i) increase the Target Redemption Amount of the Employee Vehicle above **$150 billion**\"; \"(vii) effect or permit a merger, division, or consolidation of the Partnership ... knowingly take or permit any other action to occur that would adversely affect or otherwise alter the structure of the Partnership; provided that if the General Partner reasonably determines in good faith that a value-aligned safety-conscious project (the 'Project') will within a two year period build AGI ... the General Partner can proceed (and such action shall not be treated as a Major Decision) to effect or permit a merger or consolidation of the Partnership with or into such Person ('Merger Transaction') only if such Person is running the Project and is, or is controlled by, a U.S. federal governmental entity or a non-profit entity whose board of directors is not majority controlled by Persons who are employed by ... a competitor (or its affiliate) of the Converted Second Close Limited Partner ...\"\n> - **Lead Investor:** \"the Target Redemption Amount of the Lead Investor shall be deemed to be five times (5X) its actual Target Redemption Amount\" for Majority-In-Interest calculations.\n\n> **§ 3.1 Capital Commitments.** \"In no event shall the initial Target Redemption Amount of a First Close Limited Partner be more than 100 times its Capital Commitment; provided, however, that the inflation adjustment ... will cause such Target Redemption Amounts to exceed such multiple over time.\"\n>\n> **§ 6.4 Subordination of Partnership and Limited Partner Interests to Core Mission; Modification of Fiduciary Duties** (verbatim from the all-caps recital at MSFT_MUSK000059984 / 201.034):\n>\n> > **(a) SUBJECT TO THE PROVISIONS OF THIS AGREEMENT, THE PARTNERS ACKNOWLEDGE AND AGREE THAT AS DETERMINED BY THE GENERAL PARTNER IN ITS SOLE DISCRETION, THE INTERESTS OF THE PARTNERSHIP AND THE LIMITED PARTNERS SHALL BE SUBORDINATE TO THE CORE MISSION; PROVIDED THAT THE GENERAL PARTNER AND THE GP RELATED PERSONS SHALL NOT CONSIDER THEIR OWN ECONOMIC INTERESTS BEFORE THE INTERESTS OF THE PARTNERSHIP. ... UNDER NO CIRCUMSTANCES SHALL THE GENERAL PARTNER BE LIABLE TO THE PARTNERSHIP OR THE LIMITED PARTNERS IN CONSEQUENCE OF HAVING TAKEN ANY ACTION WITHOUT MATERIAL MISCONDUCT INTENDED TO FURTHER THE CORE MISSION, EVEN IF SUCH ACTION CONFLICTS WITH THE INTERESTS OF OR, ACTUALLY RESULTS IN ADVERSE CONSEQUENCES TO, THE PARTNERSHIP OR THE LIMITED PARTNERS. ... NONE OF THE GENERAL PARTNER, ANY GP RELATED PERSON, OR ANY AFFILIATE OR ASSOCIATE OF ANY OF THE FOREGOING, SHALL BE DEEMED TO HAVE VIOLATED ANY FIDUCIARY OR OTHER DUTY TO ANY PERSON IN CONSEQUENCE OF HAVING PLACED THE CORE MISSION AHEAD OF THE INTERESTS OF THE PARTNERSHIP OR THE LIMITED PARTNERS ...\"**\n>\n> The General Partner may, \"reasonably and in good faith,\" cause the Partnership to take a list of Core-Mission actions, including:\n> > \"(xi) Retain any and all cash or property held by the Partnership as the General Partner deems necessary to fund the acquisition of any business (including one or more large publicly-traded businesses) that the General Partner reasonably and in good faith determines could achieve AGI (prior to the Declaration of Sufficient AGI), even if the manner of such achievement of AGI conflicts with the Core Mission.\"\n>\n> § 6.4(b)(i)–(iii) (the \"expressly acknowledged risks\"):\n> > \"The Core Mission is exceedingly difficult, that achievement of the Core Mission is exceedingly speculative, and that achievement of the Core Mission may not be attained. ... Only if the General Partner has affirmatively determined, in its Sole Discretion, that the Partnership is actually holding a greater amount of cash on hand than is needed to achieve the Core Mission will the General Partner begin contemplating distributions under Section 5.1(b) to the Partners.\"\n>\n> § 6.4(i) (fiduciary-duty waiver):\n> > \"Each Limited Partner hereby irrevocably: (i) waives any and all current and future claims (and rights to assert such claims) against the General Partner and any other Indemnified Person for breach of any fiduciary duty that would otherwise arise under applicable law but would be inconsistent with the terms of this Agreement ...\"\n\n> **Signature page (MSFT_MUSK000060023 / 201.073):**\n>\n> > GENERAL PARTNER: **OPENAI GP, L.L.C.**, a Delaware limited liability company\n> > By: ___ Name: **David Lansky** Title: **Manager**\n> >\n> > LIMITED PARTNERS: **MICROSOFT CORPORATION**\n> > By: ___ Name: **Satya Nadella** Title: **Chief Executive Officer**\n\n> **SCHEDULE A — Partner Information (selected entries):**\n>\n> - **OpenAI GP, L.L.C.** (General Partner) — Capital Commitment $4,778,508,000 / **Target Redemption $100 Billion** (also a $5,000,000,000 line)\n> - **OpenAI, Inc.** (Non-Profit) — Capital Commitment $47,785,080\n> - **OpenAI Holdings, L.P.** — $0 / $50,000,000\n> - **Aphorism Foundation** (Menlo Park) — c/o Zack Herlick\n> - **Gregory S. Jensen Revocable Trust** (Bridgewater's Greg Jensen) — five tranches totaling **$1,083,000,000** in Capital Commitments / **$28.3 Billion** in Target Redemption Amounts (with the $1B tranche \"subject to the conditions set forth in Section 3.2(b)\")\n> - **The Buchheit Revocable Trust** (Paul Buchheit, Gmail creator)\n> - **Khosla Ventures VI (AIV), L.P.**\n> - **YC Holdings II AIV, LLC** (Y Combinator)\n> - **Microsoft Corporation** — One Microsoft Way, Redmond, WA 98052\n\n> **EXHIBIT A — OpenAI Charter** (April 9, 2018, printed from openai.com/charter, MSFT_MUSK000060027 et seq.):\n>\n> > \"OpenAI's mission is to ensure that artificial general intelligence (AGI) — by which we mean highly autonomous systems that outperform humans at most economically valuable work — benefits all of humanity. We will attempt to directly build safe and beneficial AGI, but will also consider our mission fulfilled if our work aids others to achieve this outcome.\n> >\n> > **Broadly Distributed Benefits** — We commit to use any influence we obtain over AGI's deployment to ensure it is used for the benefit of all, and to avoid enabling uses of AI or AGI that harm humanity or unduly concentrate power. Our primary fiduciary duty is to humanity. ...\n> >\n> > **Long-Term Safety** — We are committed to doing the research required to make AGI safe ... if a value-aligned, safety-conscious project comes close to building AGI before we do, we commit to stop competing with and start assisting this project. We will work out specifics in case-by-case agreements, but a typical triggering condition might be 'a better-than-even chance of success in the next two years.'\n> >\n> > **Technical Leadership** ... **Cooperative Orientation** — We will actively cooperate with other research and policy institutions; we seek to create a global community working together to address AGI's global challenges. ...\"\n\n> **EXHIBIT B — ESCROW AND TRUST DEPOSIT AGREEMENT** (July 2, 2019), among **SunTrust Bank** (Custodian), **Microsoft Corporation**, and **OpenAI, L.P.** (\"Depositor\"), in connection with the **Joint Development and Collaboration Agreement** of the same date among Microsoft, OpenAI, L.P., and OpenAI, Inc. The escrow holds the \"Deposited Amounts\" in trust **for the exclusive benefit of Microsoft for the purpose of providing funds to Microsoft for products, services or resources provided to the Depositor by Microsoft in accordance with the Development Agreement** — i.e., the Microsoft Azure-credits-and-services flow. Custodian is to distribute funds to Microsoft from the Capital Contribution Trust Account on Microsoft's solo invoice notice; an LP Reserve Escrow Account is governed by joint Microsoft+Depositor instructions.\n\n[Pages 30–66 of the LPA: Sections 6.5–10.28 (Expenses, Partner Compensation, Records, Confidentiality, Disclosures, Holding/Feeder Vehicles, Transfers and Withdrawals, Dissolution and Liquidation, Liability and Indemnification, Meetings, Notices, Governing Law, Investment Advisers Act / Exchange Act / Bad Actor / CFTC / Anti-Corruption / Whistleblower compliance, etc.) — standard LP boilerplate, but note Section 10.13 substitutes AAA arbitration for normal accounting actions and Section 10.16 selects Delaware law. Full text in source PDF: 0201.pdf, also split as 0201.part*.pdf.]\n\n[Pages 67–93: Continued Schedule A entries, signature blocks for additional LPs, Exhibit B escrow body and exhibits (Authorized Representatives lists, fee schedules) — administrative text; full text in source PDF.]\n\n## Commentary\n\nPX 201 is the **legal foundation of the for-profit conversion that plaintiffs say betrayed the 2015 Founding Agreement**. Three doctrinal payloads explain why plaintiffs put it in on Day 7 alongside the Wu 30(b)(6) testimony (see [[Day 7]] / [[Key Themes]] § \"Day 7 — Microsoft economics\"): (1) the **all-caps § 6.4 mission-subordination + fiduciary-duty waiver**, which plaintiffs cite as proof that the new entity formally subordinated investor (and thus charitable-donor) economic interests to the Core Mission, while simultaneously authorizing the GP to \"spend, invest or otherwise use any of the Partnership's cash\" on AGI-acquisition transactions even where they \"conflict[] with the Core Mission\" (§ 6.4(a)(xi)); (2) **Microsoft as the \"Converted Second Close Limited Partner,\"** signed by Satya Nadella, with Microsoft-only Major-Decision blocks on mergers (§ 6.4 / \"Major Decisions\" (vii)) — the structural origin of plaintiffs' theory that Microsoft, not the OpenAI nonprofit, holds the practical veto over restructuring; and (3) the **Schedule A target-redemption stack** that plaintiffs aggregate to the **$250B figure** Wu was walked through — $100B for the GP/employee vehicle plus the multi-tranche LP redemptions, all driven by the \"100×\" Capital-Commitment-to-Target-Redemption ratio. The attached **Charter** (Exhibit A) is the same \"primary fiduciary duty is to humanity\" / \"stop-and-assist\" text that Brockman was cross-examined on, and the attached **SunTrust escrow** (Exhibit B) is the operational pipe that returns Microsoft's \"investment\" to Microsoft as Azure invoices — the receipts plaintiffs use against the defense's \"real money\" framing of the Microsoft deal. Cross-reference [[Brockman Journal]] for the August 2017 control-fight passages PX 201's structure formalizes, and [[PX 157]] for Sutskever and Brockman's prophecy that the company would be \"much more important than any individual\" once on track to AGI — § 6.4 is now that structure on paper.\n\n---\n*See also:* [[Day 7|Day 7 digest]] · [[Robert Wu]] · [[Greg Brockman]] · [[Brockman Journal]] · [[PX 157]] · [[Key Themes]] · [[Watershed MOU]]\n"}
{"exhibit_id": "PX-202", "exhibit": "PX 202", "party": "Plaintiffs", "type": "Contract", "admitted_trial_day": "Day 7 (May 5, 2026)", "uploaded_box_pt": "2026-05-05T16:29:57", "uploader": "Someone", "pages": 36, "size_bytes": 12326024, "source_pdf": "PX-202.pdf", "pdf_url": "https://media.mts-in.com/PX-202.pdf", "body_markdown": "# PX 202 — July 2, 2019 Microsoft / OpenAI Joint Development and Collaboration Agreement (JDCA)\n\n> The fully executed three-party Microsoft / OpenAI Inc. / OpenAI LP **Joint Development and Collaboration Agreement** dated July 2, 2019 — the foundational contract for the $750M Azure spending commitment, the GPT/Selected Technology exclusive license, the AGI carve-out, and the Microsoft Azure GPU hardware ramp.\n\n## Document type\n**Contract.** 36 pages: cover page (with later DocuSigned signature page bearing Sam Altman's June 28, 2019 signature for OpenAI Inc. and OpenAI LP; Microsoft signed by Satya Nadella), General Terms (Sections 1–13), Exhibit A (Specifications), Exhibit B (Relationship Management / Governing Board), Exhibit C (Form of SOW), Exhibit D (Trademarks). Bates MSFT_MUSK000055169–055204 (originally HIGHLY CONFIDENTIAL).\n\n## Logistics\n- **Trial admission:** Day 7 (May 5, 2026) — used during Brockman defense direct + Mr. Molo recross and the Wu deposition video. No prior wiki reference; admission day inferred from Box upload date.\n- **Box upload:** 2026-05-05 16:29:57 PT — Day 7 late-afternoon batch.\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~11.8 MB, 36 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `0202.pdf`.\n\n## Transcribed text\n\n> **JOINT DEVELOPMENT AND COLLABORATION AGREEMENT**\n>\n> Among **Microsoft Corporation**, **OpenAI, Inc.** (Delaware 501(c)(3) nonprofit), and **OpenAI, L.P.** (Delaware LP). Effective Date: **July 2, 2019**. Signed (DocuSign): **Sam Altman, CEO**, for OpenAI Inc. and for OpenAI LP (via OpenAI GP, L.L.C., its general partner; OpenAI Inc., its sole member), execution date **6/28/2019**; **Satya Nadella, CEO**, for Microsoft.\n>\n> **Background.** \"Microsoft is a worldwide provider of software and services, including its cloud-based computing platform and cloud computing services collectively known as Microsoft Azure ('Azure'). OpenAI is a non-profit entity organized under Section 501(c)(3) of the U.S. Internal Revenue Code and is a worldwide leader in the development and distribution of artificial intelligence technologies. OpenAI has established an affiliate, LP, to advance OpenAI's mission and, potentially, to commercially exploit certain of its artificial intelligence technologies. LP will be capitalized by Microsoft and other investors…\"\n>\n> **Section 1 — Definitions (selected).**\n> - **\"AGI\" or \"Artificial General Intelligence\"** — \"a highly autonomous system that outperforms humans at most economically valuable work.\"\n> - **\"Language Model Technology\"** — \"any LP or OpenAI model trained on the objective of correctly predicting words from context… and the model known as GPT-2 and all successors to and future versions thereof,\" excluding joint image/language and game-based RL (Dota / Atari). Includes Improvements developed in the Collaboration.\n> - **\"Selected Technology\"** — to be jointly agreed via the Governing Board within 90 days of effective date.\n> - **\"Dedicated Hardware\"** — clusters of GPUs and other hardware procured, hosted, and maintained by Microsoft, dedicated to LP/OpenAI use.\n> - **\"Excluded License\"** — any license requiring source disclosure / derivative-works / royalty-free redistribution (i.e., copyleft-style open source).\n>\n> **Section 2 — Collaboration and development.**\n> - 2(f) **Supercomputer development.** Parties collaborate on (i) a **Supercomputer Processing Unit** operational in CY2020 and (ii) an **AGI Relevant Supercomputer**. On Completion of the AGI Relevant Supercomputer, Microsoft makes it available exclusively to Microsoft, LP, OpenAI, and Permitted Affiliates for **two (2) calendar years**, after which Microsoft may make it freely available to third parties.\n> - 2(g) **Limited Exclusivity.** \"During the Term, LP, OpenAI and their respective Affiliates will collaborate **exclusively with Microsoft** … on designing and building any Supercomputer, Supercomputer Tech, or other supercomputing technology.\" Narrow carve-outs for Google TPU testing through March 30, 2020.\n> - 2(h) **Migration to Azure.** OpenAI/LP \"will, beginning on the Effective Date, port their cloud services… to Azure as soon as commercially practical.\"\n> - 2(j) **Open source.** Governing Board reviews any open-sourcing. OpenAI/LP make final call on Language Model Technology open-sourcing; Microsoft on Supercomputer Tech. Selected Technology may be open-sourced only after at least 12 months in Microsoft's hands and after Microsoft has license to the most advanced version.\n>\n> **Section 3 — Licenses and ownership.**\n> - 3(a) Ownership splits: Microsoft owns Supercomputer Work Product; LP owns Language Model Work Product; OpenAI owns Improvements to its own Subject Background Technology. Pooled Work Product allocated by alternating \"**Draft**\" process; Microsoft drafts first.\n> - 3(g) **Future Technology License (the \"Selected Technology\" license).** Within 90 days, parties select either a state-of-the-art language model or other technology. LP and OpenAI grant Microsoft \"an **exclusive** (except for LP's and OpenAI's non-commercial research use and as permitted pursuant to Section 2(j)), perpetual, irrevocable, worldwide, royalty-free, fully paid\" license to make, use, sell, distribute, and create derivative works of the Selected Technology in combination with Microsoft Licensed Products. **OpenAI has the final decision** but \"will not unreasonably withhold or delay agreement to include any such language model or alternative technology that passes the safety review for Microsoft's intended use case.\"\n> - 3(h) **Data rights.** Worldwide, nonexclusive, perpetual, royalty-free cross-license to \"access, copy, and use any data provided by that party,\" limited by Applicable Technology category.\n>\n> **Section 4 — Right of first negotiation.** \"Throughout the Term, if LP or OpenAI intends to make any Commercial Use of any LP or OpenAI Technology other than the Selected Technology **or an AGI system**, LP and OpenAI will first notify Microsoft … and the parties will then negotiate, exclusively and in good faith, the terms of an exclusive license to Microsoft.\" (60-day exclusive negotiation window.)\n>\n> **Section 5 — Commercial terms.**\n> - 5(a) **Azure hardware commitment** (Table 1):\n>   - At least **10,000** V100-equivalent GPUs by **January 30, 2020**\n>   - At least **25,000** by December 31, 2020\n>   - At least **50,000** by December 31, 2021\n> - 5(b) **Spending commitment.** \"LP and OpenAI will, in the aggregate, after the Effective Date, spend at least **seven hundred fifty million US dollars ($750,000,000)** on Azure services… no later than December 31, 2024 or five (5) years after Microsoft has made the Improved Azure Feature Set available.\"\n> - 5(c) **Escrow.** Capital Contribution Trust Account funded by LP from Microsoft Investment Agreement proceeds; disbursed to Microsoft on invoice or termination.\n> - 5(d) **Pricing model.** 55% discount on Consumption Rates for generally available Azure services; cost-plus model for Dedicated Hardware (specific rates redacted).\n>\n> **Section 6 — Other rights and obligations.**\n> - 6(a) **Governance** by a Governing Board co-chaired by Sam Altman (LP/OpenAI) and Kevin Scott (Microsoft EVP, CTO). Initial members: Altman, Mira Murati, Christopher Berner (LP/OpenAI side); Scott, Girish Bablani (CVP Azure Compute), Eric Boyd (CVP) (Microsoft side). Unanimous decisions; deadlock → Section 2 Dispute resolution.\n> - 6(c) **Safety review.** OpenAI and LP develop and share Safety Criteria. \"LP and OpenAI will conduct a thorough and prompt safety review of each version of the Language Model Technology … intended for commercial or research use by Microsoft.\" OpenAI/LP make final call. Approval not revocable without Governing Board agreement.\n>\n> **Section 12 — Term and termination.**\n> - 12(a) Term to December 31, 2024 (or Spending Deadline if later); one optional 2-year extension on notice by June 30, 2024.\n> - 12(b)(iii) **Change Event termination right (Microsoft only)** on 180 days' notice if (A) regulators in the US, EU/UK, or Canada deem the AI illegal/substantially regulated, or (B) \"**a change in the Chief Executive Officer of OpenAI**.\"\n>\n> **Section 13(b) — Governing law.** New York law; SDNY federal jurisdiction or NY County state courts.\n>\n> **Exhibit A — Specifications.**\n> - GPU Servers: 32GB V100-equivalent GPUs with NVLink, ≥8 GPUs/server, 100Gbit/s NIC, RAM ≥2.5x GPU memory.\n> - Network: 32 Gbit/s, 16 Gbit/s per TCP stream, scalable to 8k VMs (2019), 16k (2020); 100 Gbit/s GPU interconnect scalable to **8000 V100-equivalent GPUs in 2020, 16k in 2021**.\n> - **AGI Relevant Supercomputer description:** \"**a system that can train a one trillion parameter sparse model.**\"\n> - Supercomputer Processing Unit description: \"the functional equivalent or better than TPU v3 pods.\"\n>\n> **Exhibit B — Governing Board** (Table B1):\n> | LP / OpenAI | Microsoft |\n> |---|---|\n> | Sam Altman, CEO (Co-Chair) | Kevin Scott, EVP, CTO (Co-Chair) |\n> | Mira Murati, MTS | Girish Bablani, CVP Azure Compute |\n> | Christopher Berner, MTS | Eric Boyd, Corporate VP |\n>\n> **Exhibit D — Trademarks.** Microsoft Azure logo licensed; OpenAI logo (U.S. Trademark Reg. **5,448,123**) licensed.\n>\n> [Document also includes the customary Confidentiality (Section 8 / NDA reference), Disclaimers (Section 10 — $200K direct / $1M aggregate liability cap with carve-outs), Defense of Third-Party Claims (Section 11 — joint and several for LP and OpenAI), and General provisions (Section 13). Full verbatim text is in the source PDF.]\n\n## Commentary\n\nPX 202 is the **single most consequential corporate document in the case after the [[PX 24]] Charter and the [[PX 236|2018 LP term sheet]].** Plaintiffs use it to make four distinct points: (i) the 2019 contract memorialized OpenAI's representation to Microsoft that **OpenAI Inc. is \"a non-profit entity organized under Section 501(c)(3)\"** and that the LP exists \"to advance OpenAI's mission\" — not to displace it; (ii) Section 3(g)'s **exclusive, royalty-free, perpetual** license of the Selected Technology to Microsoft is plaintiffs' Exhibit A for the \"**OPEN means open-source betrayal**\" theme — it precludes broad open-sourcing for at least 12 months and conditions any release on Microsoft having a more-advanced version (see [[Key Themes#The \"OPEN\" in OpenAI]]); (iii) Section 4's right-of-first-negotiation **carves out \"an AGI system\"** — confirming the parties' contemporaneous understanding that AGI sits **outside** the Microsoft commercial deal, the predicate for the much-debated \"AGI clause\" in later iterations; and (iv) Section 12(b)(iii)(B) — Microsoft's right to terminate on **a change in OpenAI's CEO** — is the contractual hook plaintiffs and the Toner deposition use to explain Microsoft's rapid Sutskever-board-side intervention during the [[Day 8|November 2023 board crisis]]. The $750M Azure commitment + 50,000-GPU ramp + 55% discount + capital-contribution escrow trace the [[Microsoft thread]] from the [[PX 90|June 2017 \"10,000 servers\"]] ask to the post-2019 Azure dependency. Defense will use the same document for an opposite reading: that the 2019 deal was disclosed, board-approved, and structurally identical to what Musk himself contemplated in [[PX 25]] (\"**a standard C corp with a parallel nonprofit**\").\n\n---\n*See also:* [[Day 7|Day 7 digest]] · [[Sam Altman]] · [[Greg Brockman]] · [[PX 24]] · [[PX 236]] · [[PX 90]] · [[Microsoft thread|Key Themes — Microsoft thread]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-203", "exhibit": "PX 203", "party": "Plaintiffs", "type": "Contract", "admitted_trial_day": "Day 7 (May 5, 2026)", "uploaded_box_pt": "2026-05-05T17:54:59", "uploader": "Someone", "pages": 112, "size_bytes": 12018642, "source_pdf": "PX-203.pdf", "pdf_url": "https://media.mts-in.com/PX-203.pdf", "body_markdown": "# PX 203 — Second Amended and Restated Limited Partnership Agreement of OpenAI, L.P. (March 6, 2021), with OpenAI Charter and SunTrust escrow agreement\n\n> The full executed text of the OpenAI, L.P. limited partnership agreement — the \"purple box\" donation-in-spirit warning, the AGI carveout, the Microsoft \"Convertible Limited Partnership Interest\" with a Maturity Date of July 2, 2049, and the schedule of partners' Capital Commitments and Target Redemption Amounts (Microsoft $1B/$20B and $2B/$12B; nonprofit $0/$100B; Aphorism $50M/$5B; Khosla $50M/$5B; YC $10M/$1B; Buchheit $3M/$300M; Jensen $20M/$2B).\n\n## Document type\n**Contract.** A 78-page Delaware limited-partnership agreement plus signature page, Schedule A (Partner Information), Exhibit A (OpenAI, Inc. Charter), and Exhibit B (SunTrust Escrow and Trust Deposit Agreement, July 2, 2019, with March 6, 2021 Amendment No. 1). Signed by Sam Altman as Chief Executive Officer of OpenAI GP, L.L.C. (general partner) and as attorney-in-fact for the Limited Partners; David Lansky and Brad Lightcap are listed as authorized representatives. Bates: MSFT_MUSK000064573–64684.\n\n## Logistics\n- **Trial admission:** Day 7 (May 5, 2026) — Brockman defense direct + Mr. Molo recross + DX 1252; Wu deposition. PX 203 is the underlying contract authenticated through Robert Wu's 30(b)(6) Microsoft deposition; the figures Wu walked the jury through (the **$20B target redemption stack**, the **20% annual escalator**, the **AGI carveout** in § 6.4) live in this document. See [[Key Themes]] § \"Day 7 — the Microsoft economics.\"\n- **Box upload:** 2026-05-05 17:54:59 PT — uploaded same day as Wu's deposition testimony.\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~12.0 MB, 112 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `0203.pdf`.\n\n## Transcribed text\n\n> **SECOND AMENDED AND RESTATED LIMITED PARTNERSHIP AGREEMENT OF OPENAI, L.P. — A DELAWARE LIMITED PARTNERSHIP — March 6, 2021**\n\n### The \"purple box\" — Section preamble\n\n> **IMPORTANT**\n> ***Investing in OpenAI, L.P. is a high-risk investment***\n> ***Investors could lose their capital contribution and not see any return***\n> ***It would be wise to view any investment in OpenAI, L.P. in the spirit of a donation, with the understanding that it may be difficult to know what role money will play in a post-AGI world***\n>\n> The Partnership exists to advance OpenAI, Inc.'s mission of ensuring that safe artificial general intelligence is developed and benefits all of humanity. The General Partner's duty to this mission and the principles advanced in the OpenAI, Inc. Charter take precedence over any obligation to generate a profit. The Partnership may never make a profit, and the General Partner is under no obligation to do so. The General Partner is free to re-invest any or all of the Partnership's cash flow into research and development activities and/or related expenses without any obligation to the Limited Partners. See Section 6.4 for additional details.\n\n### Section 2.3 — Purpose and Scope (the \"Core Mission\")\n\n> Within the meaning and for purposes of the Act, the purpose and scope of the Partnership shall be to:\n> (a) Use the principles expounded in the Non-Profit Charter to advance the Non-Profit's mission of pursuing (i) the creation, development and refinement of a safe artificial general intelligence — meaning a highly autonomous system that outperforms humans at most economically valuable work — (\"**AGI**\"); and (ii) the dissemination and distribution of the benefits of such AGI among all of humanity (collectively, the \"**Core Mission**\").\n> (b) Subordinate to the Core Mission, seek income and gain through selective licensing, commercialization or other monetization of technologies, scientific breakthroughs or other similar discoveries that are achieved in the pursuit, or furtherance, of the Core Mission;\n> (c) Temporarily invest cash in Idle Funds Investments; and\n> (d) Engage in any other lawful activities (including activities described in Section 6.2) determined by the General Partner to be necessary or advisable in furtherance of the foregoing activities.\n\n### Key defined terms\n\n> **Converted Limited Partner** means **Microsoft Corporation**.\n>\n> **Convertible Limited Partnership Interest** … (i) the holder of which may convert to a Second/Third Close Limited Partner on or after **July 2, 2026** and the holder of which shall convert to a Second/Third Close Limited Partner on the date that is no later than **July 2, 2049** (\"**Maturity Date**\"), with such conversion date subject to (1) an acceleration of the then applicable conversion date upon the **Declaration of Sufficient AGI**, and (2) extension by the Converted Second/Third Close Limited Partner in its sole discretion …\n>\n> **Declaration of Sufficient AGI** shall mean the effective time of the General Partner's determination, in its reasonable discretion, that (x) AGI has been created that has the capability to generate the Target Redemption Amount for each Partner; and (y) it has the ability and authority to direct the AGI to generate such Target Redemption Amounts.\n>\n> **Lead Investor** shall mean the **Aphorism Foundation**.\n>\n> **Second Closing Capital Commitment** shall mean **one billion dollars ($1,000,000,000)**.\n>\n> **Third Closing Capital Commitment** shall mean **two billion dollars ($2,000,000,000)**.\n\n### Section 6.4 — Subordination of Partnership and Limited Partner Interests to Core Mission (the \"AGI carveout\")\n\n> (a) SUBJECT TO THE PROVISIONS OF THIS AGREEMENT, THE PARTNERS ACKNOWLEDGE AND AGREE THAT AS DETERMINED BY THE GENERAL PARTNER IN ITS SOLE DISCRETION, **THE INTERESTS OF THE PARTNERSHIP AND THE LIMITED PARTNERS SHALL BE SUBORDINATE TO THE CORE MISSION**; PROVIDED THAT THE GENERAL PARTNER AND THE GP RELATED PERSONS SHALL NOT CONSIDER THEIR OWN ECONOMIC INTERESTS BEFORE THE INTERESTS OF THE PARTNERSHIP. … UNDER NO CIRCUMSTANCES SHALL THE GENERAL PARTNER BE LIABLE TO THE PARTNERSHIP OR THE LIMITED PARTNERS IN CONSEQUENCE OF HAVING TAKEN ANY ACTION WITHOUT MATERIAL MISCONDUCT INTENDED TO FURTHER THE CORE MISSION, EVEN IF SUCH ACTION CONFLICTS WITH THE INTERESTS OF OR, ACTUALLY RESULTS IN ADVERSE CONSEQUENCES TO, THE PARTNERSHIP OR THE LIMITED PARTNERS. …\n>\n> (b)(ii) **Only if the General Partner has affirmatively determined, in its Sole Discretion, that the Partnership is actually holding a greater amount of cash on hand than is needed to achieve the Core Mission will the General Partner begin contemplating distributions under Section 5.1(b) to the Partners.**\n\n### Section 5.1(b)(ii) — Distribution waterfall after Commencement of the Redemption Period\n\n> (A) First, to the First Close Limited Partners (including the Non-Profit with respect to its Limited Partner Investment Interest) until each such Partner has received aggregate distributions equal to its Capital Contributions.\n> (B) Next, any remaining items of cash or property … shall be apportioned and distributed (1) **25 percent**, in proportion to each Partner's Target Redemption Amount, to the First Close Limited Partners …, and to the Employee Vehicle; and (2) **75 percent**, in proportion to their respective amounts remaining to be distributed pursuant to this Section 5.1(b)(ii)(B), to the Second/Third Close Limited Partners, until each such Second/Third Close Limited Partner has received aggregate distributions equal to its Capital Contributions.\n> (C) Next, … until the time that each has received aggregate distributions … equal to its respective Target Redemption Amount …\n> (D) Next, all remaining items of cash and property shall be **distributed to the Non-Profit in its capacity as a Limited Partner**.\n\n### Schedule A — Partner Information\n\n| Limited Partner | Capital Commitment | Target Redemption Amount |\n|---|---:|---:|\n| **OpenAI, Inc.** (Non-Profit) | $60,829,083 | $6,082,908,300 |\n| **OpenAI Holdings, L.P.** (Employee Vehicle) | $0 | **$100 Billion** |\n| **Aphorism Foundation** | $50,000,000 | $5,000,000,000 |\n| **Gregory S. Jensen Revocable Trust** | $20,000,000 | $2,000,000,000 |\n| **The Buchheit Revocable Trust** | $3,000,000 | $300,000,000 |\n| **Khosla Ventures VI (AIV), L.P.** | $50,000,000 | $5,000,000,000 |\n| **YC Holdings II AIV, LLC** | $10,000,000 | $1,000,000,000 |\n| **Microsoft Corporation** (Second Closing) | $1,000,000,000\\* | **$20,000,000,000** |\n| **Microsoft Corporation** (Third Closing) | $2,000,000,000\\*\\* | **$12,000,000,000** |\n\n\\* Second Closing Capital Commitment; subject to the conditions set forth in Section 3.2(b)(i). \\*\\* Third Closing Capital Commitment; subject to the conditions set forth in Section 3.2(b)(ii).\n\n### Target Redemption Amount escalator\n\n> The Target Redemption Amount … shall be increased by … (x) for the first ten years occurring after the Initial Closing, … the rate of inflation … (the \"Inflation Rate\"); and (y) **for each full or partial Fiscal Year that begins after the tenth anniversary of the Initial Closing, the Target Redemption Amount for each Partner shall be increased by the greater of the Inflation Rate for such full or partial Fiscal Year, and a percent equal to LIBOR plus 2 percent.**\n\n### Exhibit A — OpenAI, Inc. Charter (verbatim)\n\n> OpenAI's mission is to ensure that artificial general intelligence (AGI) — by which we mean highly autonomous systems that outperform humans at most economically valuable work — benefits all of humanity. We will attempt to directly build safe and beneficial AGI, but will also consider our mission fulfilled if our work aids others to achieve this outcome. To that end, we commit to the following principles:\n>\n> **Broadly Distributed Benefits.** We commit to use any influence we obtain over AGI's deployment to ensure it is used for the benefit of all, and to avoid enabling uses of AI or AGI that harm humanity or unduly concentrate power. **Our primary fiduciary duty is to humanity.** We anticipate needing to marshal substantial resources to fulfill our mission, but will always diligently act to minimize conflicts of interest among our employees and stakeholders that could compromise broad benefit.\n>\n> **Long-Term Safety.** We are committed to doing the research required to make AGI safe, and to driving the broad adoption of such research across the AI community. We are concerned about late-stage AGI development becoming a competitive race without time for adequate safety precautions. **Therefore, if a value-aligned, safety-conscious project comes close to building AGI before we do, we commit to stop competing with and start assisting this project.** We will work out specifics in case-by-case agreements, but a typical triggering condition might be \"a better-than-even chance of success in the next two years.\"\n>\n> **Technical Leadership.** To be effective at addressing AGI's impact on society, OpenAI must be on the cutting edge of AI capabilities — policy and safety advocacy alone would be insufficient. We believe that AI will have broad societal impact before AGI, and we'll strive to lead in those areas that are directly aligned with our mission and expertise.\n>\n> **Cooperative Orientation.** We will actively cooperate with other research and policy institutions; we seek to create a global community working together to address AGI's global challenges. We are committed to providing public goods that help society navigate the path to AGI. Today this includes publishing most of our AI research, but we expect that safety and security concerns will reduce our traditional publishing in the future, while increasing the importance of sharing safety, policy, and standards research.\n\n### Exhibit B — Escrow and Trust Deposit Agreement (July 2, 2019; Amendment No. 1, March 6, 2021)\n\n> This Agreement … by and among **SunTrust Bank**, a Georgia banking corporation (the \"Custodian\"), **Microsoft Corporation** … and **OpenAI, L.P.** … in connection with that certain Joint Development and Collaboration Agreement, dated as of July 2, 2019 … and that certain Amended and Restated Limited Partnership Agreement of the Depositor dated as of July 2, 2019 …\n>\n> 1.1.1 … the \"**Capital Contribution Trust Account**\" … the \"**LP Reserve Escrow Account**\" …\n> 1.1.2 The **Capital Contribution Trust Amount** will be held as a trust fund for the **exclusive benefit of Microsoft** for the purpose of providing funds to Microsoft for products, services or resources provided to the Depositor by Microsoft in accordance with the Development Agreement. … the Depositor shall not retain any legal or equitable interest in the Capital Contribution Trust Account and any **Unilateral Instruction will not require the consent or signature of Depositor**.\n\nSignatures: Keith Dolliver (Microsoft, Assistant Secretary), David Lansky (OpenAI GP, L.L.C., Manager), Matt Ward / Charles Henderson (SunTrust/Truist, Custodian).\n\n[Sections 1, 4 (Profits and Losses), 7 (Transfers), 8 (Dissolution), 9 (Liability and Indemnification), and 10 (General Provisions, including the Morrison & Foerster legal-counsel waiver in § 10.18) read in the document substantially as transcribed in the PDF and are quoted in full at Bates MSFT_MUSK000064579–64656; the foregoing extracts are the ones actually deployed at trial.]\n\n## Commentary\n\nPX 203 is the legal architecture of the OpenAI, L.P. for-profit — the document plaintiffs use to demonstrate that the for-profit's \"fiduciary duty\" to its mission is structurally subordinated to the Core Mission only on paper, while the actual cash waterfall is a stack of redemption preferences capped at Target Redemption Amounts that escalate at LIBOR + 2% per year. **Plaintiffs' read:** the Schedule A figures are the receipts — Microsoft's $3B in commitments map to a **$32B target redemption**, OpenAI Holdings (the \"Employee Vehicle\") starts at **$100 billion** in target redemption with no cash in, and the Non-Profit's $60.8M Limited Partner Investment Interest sits with a $6.08B target — a 100x cap that sets up the \"**zero**\" Brockman testimony in [[Key Themes]] and the \"$30 billion / no money in\" Molo set piece. **Defense's read:** the **purple box** (\"in the spirit of a donation … may never make a profit\") and the **AGI carveout** in § 6.4 are exactly the limiting language Musk ignored. Wu's Day 7 testimony — the [[Key Themes]] § \"Day 7 — the Microsoft economics\" — walks the jury through these exact provisions: the **20% annual escalator**, the **6× return target on the 2023 Microsoft round**, the **Watershed MOU** AGI-carveout dilution, and the **uncapped-stock conversion** that the mission strategy committee began considering in April–May 2024 (see Wu @ ~15:11–15:20 PT, [[Day 7|Day 7 digest]]). The Charter's \"**Our primary fiduciary duty is to humanity**\" line in Exhibit A is the one Brockman drafted out of \"Charter 2.0\" in December 2023 (PX 359, see [[Brockman Journal]] and [[Key Themes]] § \"Charter 2.0\").\n\n---\n*See also:* [[Day 7|Day 7 digest]] · [[Robert Wu]] · [[Key Themes]] · [[PX 359]] · [[Greg Brockman]] · [[Brockman Journal]]\n"}
{"exhibit_id": "PX-204", "exhibit": "PX 204", "party": "Plaintiffs", "type": "Commercial agreement — Amended and Restated Joint Development and Collaboration Agreement (with Exhibits A–N)", "admitted_trial_day": "Day 7 (May 5, 2026)", "uploaded_box_pt": "2026-05-05 (Day 7 batch)", "uploader": "Someone", "pages": 64, "size_bytes": 22020096, "source_pdf": "PX-204.pdf", "split_parts": "0204.part1of2.pdf, 0204.part2of2.pdf", "bates": "OPENAI_MUSK00010268–OPENAI_MUSK00010331 (approx.)", "pdf_url": "https://media.mts-in.com/PX-204.pdf", "body_markdown": "# PX 204 — 2021 Amended and Restated Joint Development and Collaboration Agreement (Microsoft / OpenAI / OpenAI, L.P.)\n\n> The March 5, 2021 \"Amended JDCA\" — the master commercial contract that amends the original 2019 JDCA, executed by Satya Nadella for Microsoft and Sam Altman for OpenAI, Inc. and OpenAI, L.P. Contains the **\"Sufficient AGI\" carve-out** (§ 4(i)), the GitHub/Codex exclusivity period, the Phase 2/3/4 supercomputer roadmap, and the Deployment Safety Board's Safety-Critical thresholds.\n\n## Document type\n**Commercial agreement — multi-party master contract.** Bates-stamped OpenAI production. Includes the Cover Pages with DocuSign signature blocks; General Terms (Sections 1–12); and Exhibits A (Specifications / Phase 2–4 Supercomputers), B (Relationship Management / Governing Board / Dispute Process), C (Trademarks), D (Additional Data and Privacy Terms), E (Deployment Safety Board), F (Support and Deliverables), G (Designated Engineers), H (Trusted Engineers), I (Pricing and Payment), J (Transition — including SPEA, Bing, GPT-3 legacy terms, GitHub DPA), K (Related IP), L (Listed Countries — 26 countries including USA), M (OpenAI API Deployment Plan), and N (HPE Data Privacy and Security Agreement Schedule).\n\n## Logistics\n- **Trial admission:** Day 7 (May 5, 2026). Used during the Robert Wu 30(b)(6) deposition designations on Microsoft economics — the JDCA is the substantive flip-side of the LPA in PX 201 and the contract that defines what Microsoft actually gets for its capital-contribution-as-Azure-credits flow.\n- **Box upload:** 2026-05-05 PT — Day 7 batch.\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~21.0 MB, 64 pages. **Note:** original PDF exceeded the 20 MB Read tool cap and was split into two parts in `Elon Musk, et al. v. Samuel Altman, et al - split/`.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `0204.pdf`.\n\n## Transcribed text\n\n> **AMENDED AND RESTATED JOINT DEVELOPMENT AND COLLABORATION AGREEMENT**\n>\n> This Amended and Restated Joint Development and Collaboration Agreement (\"Agreement\") is among **Microsoft Corporation** ... (\"Microsoft\"); **OpenAI, Inc.**, a Delaware corporation ... (\"OpenAI\"); and **OpenAI, L.P.**, a Delaware limited partnership ... (\"LP\"). This Agreement is effective on **March 5, 2021** (\"Effective Date\") and it amends, supersedes, and replaces in its entirety that certain Joint Development and Collaboration Agreement entered into between the parties as of **June 28, 2019**, as amended (\"Original Agreement\") ...\n>\n> **Contact details (Table 1):**\n> - OpenAI / OpenAI, L.P.: 3180 18th Street, Suite 100, San Francisco, CA 94110; **Attn: Sam Altman, CEO**; sama@openai.com\n> - Microsoft: One Microsoft Way, Redmond, WA 98052; **Attn: Kevin Scott, CTO**\n> - General-counsel notice copies: David Lansky (OpenAI) / Deputy General Counsel, T+R (Microsoft, fax (425) 936-7329)\n>\n> **Signatures:**\n> - Microsoft: **Satya Nadella, Chief Executive Officer** — DocuSigned 3/6/2021\n> - OpenAI, Inc.: **Sam Altman, Chief Executive Officer** — March 5, 2021\n> - OpenAI, L.P.: **Sam Altman, Chief Executive Officer** — March 5, 2021\n\n> **Background recitals:**\n> \"Microsoft is a worldwide provider of software and services, including its cloud-based computing platform and cloud computing services collectively known as **Microsoft Azure** ('Azure'). OpenAI is a non-profit entity organized under Section 501(c)(3) ... and is a worldwide leader in the development and distribution of artificial intelligence technologies. OpenAI has established an affiliate, LP, to advance OpenAI's mission and to commercially exploit certain of its artificial intelligence technologies. **LP was capitalized by Microsoft and other investors**, and Microsoft, OpenAI, LP, and other investors have also entered into that certain **Amended and Restated Limited Partnership Agreement of the LP, dated July 2, 2019** ('First Investment Agreement') to facilitate such investment by Microsoft. ... The parties now wish to modify their existing agreement ... providing for **much broader licensing of IP** (defined below) to Microsoft and **much deeper engagement between the parties' respective engineering teams** to both further the mission of OpenAI and better enable Microsoft to commercialize such licensed IP ...\"\n>\n> The agreement is entered \"in consideration of the parties' entry into that certain **Second Amended and Restated Limited Partnership of Agreement of OpenAI, L.P. formed contemporaneously with this Agreement** ('Second Investment Agreement' and, collectively with the First Investment Agreement, the 'Investment Agreements').\"\n\n> **§ 1 Definitions (selected):**\n> - **\"AGI\" or \"Artificial General Intelligence\"** means a highly autonomous system that outperforms humans at most economically valuable work. For clarity, AGI excludes: (i) any systems specific to limited domains; (ii) any non-AGI specific software/middleware (e.g., Singularity, ONNX, ONNX RT, or future derivatives of or successors to any of these); and (iii) the hardware used to develop, train, host, or operate any systems.\n> - **\"Licensed IP\"** means all (i) Trained Models of LP, OpenAI, or both ... in object code form. **For clarity, Licensed IP excludes: (x) any Research; (y) all Related IP; and (z) AGI.**\n> - **\"Phase 2 / Phase 3 / Phase 4 Supercomputer\"** — the staged Azure-hosted supercomputers described in Sections 1–3 of Exhibit A.\n\n> **§ 2 Development.** \"During the Term, Microsoft will work together with LP, OpenAI, or both to Jointly Develop Technology or IP ...\" § 2(a) (Development License) grants LP/OpenAI a perpetual, irrevocable, worldwide, non-exclusive, royalty-free, fully-paid license under Microsoft's Background IP for development purposes. § 2(c)–(d) sets up the **Pooled Patents Draft** process — LP/OpenAI is treated as a single party and the parties alternately \"draft\" jointly developed patents.\n\n> **§ 3 Other IP Rights and Obligations.** § 3(b)–(c) grants Microsoft a **perpetual, irrevocable, worldwide, royalty-free, fully-paid license under each grantor's Patents and Non-Patent IP in the Licensed IP** to make, sell, distribute Microsoft Licensed Products incorporating the Licensed IP, including sublicense rights through multiple tiers — but **LP/OpenAI retains all rights to Research, Related IP, and AGI**.\n\n> **§ 4(i) Artificial General Intelligence (the \"Sufficient AGI\" provision):**\n>\n> > **\"(i) Achievement.** OpenAI may during the Term, in good faith, declare certain Models developed by LP or OpenAI to be **'Sufficient AGI'**, meaning that such Models: (A) are within the scope of the definition of AGI; and (B) are **clearly capable of generating the Target Redemption Amount (defined in the Second Investment Agreement) of return on investment capital for each of OpenAI's limited partners (including Microsoft)**, which such determination will be made pursuant to standards that **Microsoft has reviewed and approved**. On achievement of Sufficient AGI as provided in this Section 4(i)(i), LP and OpenAI may collaborate with Microsoft to monetize Sufficient AGI **until LP and OpenAI generate and pay over to each limited partner the full Target Redemption Amount** payable to each such limited partner.\n> >\n> > **(ii) Future Developments.** After LP and OpenAI pay over to each such limited partner the full applicable Target Redemption Amount in accordance with Section 4(i)(i): (A) LP, OpenAI, and Microsoft will collaborate in good faith to make the benefits of artificial general intelligence available to all humanity ('**AGI Distribution**'); and (B) **Microsoft will have the right of first refusal to be LP's and OpenAI's commercialization partner for AGI** if LP or OpenAI chooses to have a commercialization partner ... The parties will, therefore, beginning promptly on or after the achievement of AGI, enter into discussions to negotiate a further amendment to this Agreement to facilitate such Future Developments ... **No party will have any obligation whatsoever with regard to any such Future Developments unless and until all parties agree on the terms** applicable to such Future Developments and enter into a signed writing ...\"\n>\n> Termination effect: § 11(b)(iv) provides that on termination, \"**Microsoft's rights and privileges and OpenAI's and LP's obligations under Section 4(i) (Artificial General Intelligence) with regard to AGI and Sufficient AGI will be deemed removed in their entirety on a prospective basis and not replaced.**\"\n\n> **Exhibit E — Deployment Safety Board (DSB) — Safety-Critical Thresholds:**\n> \"(i) **Safety-Critical Parameters Threshold.** The AI system utilizes as a component a neural network that uses **over 1 trillion parameters in a forward inference pass**. This excludes transformer networks with mixture-of-experts layers where the total parameter count exceeds 1 trillion, but fewer than 1 trillion parameters are used in forward pass due to the 'gates' in the mixture-of-experts layers.\n> (ii) **Safety-Critical Compute Threshold.** **More than 20,000 Pf/s-days of compute** will be expended to train a neural network component of the AI system, where the amount of compute is calculated as in **Kaplan et al, 2020** (https://arxiv.org/abs/2001.08361).\"\n\n> **Exhibit J — Transition (GitHub / GPT-3 legacy terms):**\n> Re GitHub: \"The licenses granted in Sections 3(b) and 3(c) of the Agreement are and **will be exclusive during the Term with respect to Licensed IP trained on source code and related artifacts** (such as commits, pull requests, issue tickets, and data) licensed by GitHub whether licensed under the SPEA or under the Agreement ('**Subject Models**') ... The foregoing provision will apply **until the sixth anniversary of the Effective Date**, regardless of whether the Agreement is terminated or expires prior to such date. ... during such exclusivity period **LP and OpenAI may not independently develop and commercialize the OpenAI API** (or any successor versions to the OpenAI API) to the extent the OpenAI API (or successor version) consists of, uses, or relies on Subject Models.\" (i.e., a six-year Microsoft-exclusive lockup on Codex/Copilot-class models trained on GitHub data.)\n>\n> Re GPT-3: LP/OpenAI grant Microsoft an \"**exclusive** (except for LP's and OpenAI's non-commercial research use ... LP's and OpenAI's commercial use that makes the Selected Technology available via application programming interface ('API') to commercial customers), **perpetual, irrevocable, worldwide, royalty-free, fully paid**\" license to GPT-3 under Patents and Non-Patent IP, with sublicense rights to third parties — and **OpenAI agrees to make GPT-3 available to commercial customers via API solely on and through Microsoft's Azure platform** \"provided that Microsoft continues to make the Azure platform available to LP and OpenAI for such purposes on substantially the same terms as of September 22, 2020 or other commercially reasonable terms.\"\n\n> **Exhibit L — Listed Countries** (the geographies where parties may transact directly with sub-Affiliates): Australia, Austria, Brazil, Canada, Chile, Czech Republic, Denmark, Egypt, France, Germany, Greece, India, Italy, Japan, Malta, Mexico, New Zealand, Norway, Poland, Serbia, South Africa, South Korea, Spain, Sweden, United Kingdom, United States. (China is conspicuously absent — § 9 separately addresses PRC Affiliates' restricted access to Microsoft's internally-hosted OpenAI API.)\n\n> **Exhibit M — OpenAI API Deployment Plan:** \"The parties will, within 60 days after the Effective Date, enter into an amendment to include a detailed plan to develop and deploy an Azure-class successor to the OpenAI API.\"\n\n[Pages 8–40: General Terms §§ 5 (Relationship Management; Executive Sponsors; DSB; Designated/Trusted Engineers), 6 (Publicity / Joint Marketing), 7 (Confidentiality / NDA), 8 (Reps & Warranties — incl. no Excluded Licenses on Made Available IP), 9 (Compliance / China carve-outs / export), 10 (Defense of Third-Party Claims; indemnification), 11 (Term — initial term plus renewal), and 12 (Notices / Governing Law / Interpretation). Full text in source PDF: 0204.pdf, also split as 0204.part*.pdf.]\n\n[Pages 41–64: Exhibits A (Phase 2/3/4 supercomputer specs), B (Governing Board, Trigger Date, Dispute process), C (Trademarks/approver list), D (Privacy/Data terms), F (Support — Microsoft 24x7 obligations and HPE handoff for OpenAI Professional Services Data tickets), G/H (engineer rosters), I (Pricing/Payment incl. revenue-share placeholder for OpenAI API and other co-commercialized products), and N (HPE Data Privacy and Security Agreement). Full text in source PDF.]\n\n## Commentary\n\nPX 204 is **the contract** — paired with PX 201 (the LPA) — that defines plaintiffs' \"Microsoft was buying the path to AGI\" theory. Three load-bearing pieces from this exhibit show up across the Day 7 record (see [[Day 7]] / [[Key Themes]] § \"Day 7 — Microsoft economics\"): (1) **the \"Sufficient AGI\" provision in § 4(i)(i)** — AGI is \"achieved\" not by capability but by Models that are \"**clearly capable of generating the Target Redemption Amount** ... for each of OpenAI's limited partners (including Microsoft),\" with that determination made on **standards Microsoft has reviewed and approved**. Plaintiffs hammer this as proof that the AGI definition under the contract is **a financial trigger keyed to paying Microsoft and the other LPs back at 100× their committed capital** — not a safety or capability trigger — turning OpenAI's nonprofit Charter into a debt-style payout schedule. (2) **The \"AGI carve-out\" from Licensed IP** combined with the § 11 termination snap-back: AGI is excluded from Microsoft's perpetual royalty-free license, but the **Future Developments** clause gives Microsoft a **right of first refusal** to be the AGI commercialization partner — a structural option Wu was walked through. (3) **The six-year GitHub/Codex exclusivity and the GPT-3-on-Azure-only commitment** explain why, even before any new \"Watershed MOU\" deal, Microsoft already has an effective lockup on the most commercially valuable OpenAI models. The contract is signed by **Sam Altman in both his OpenAI Inc. (nonprofit) and OpenAI, L.P. (for-profit) capacities** — the structural conflict-of-interest plaintiffs use to argue that the same actor is on both sides of the donor-betrayal transaction. Cross-reference [[PX 201]] (the underlying LPA), [[Brockman Journal]] (the same dual-CEO concerns recorded in 2017), and [[PX 157]] for Sutskever and Brockman's prophecy that \"the company will be much more important than any individual\" — § 4(i) is the paper version of that prophecy.\n\n---\n*See also:* [[Day 7|Day 7 digest]] · [[PX 201]] · [[Robert Wu]] · [[Watershed MOU]] · [[Greg Brockman]] · [[Brockman Journal]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-206", "exhibit": "PX 206", "party": "Plaintiffs", "type": "Corporate governance document — Amended and Restated Limited Liability Company Agreement of OpenAI Global, LLC (with Charter and Reorganization Steps exhibits)", "admitted_trial_day": "Day 7 (May 5, 2026)", "uploaded_box_pt": "2026-05-05 (Day 7 batch)", "uploader": "Someone", "pages": 91, "size_bytes": 30408704, "source_pdf": "PX-206.pdf", "split_parts": "0206.part1of3.pdf, 0206.part2of3.pdf, 0206.part3of3.pdf", "bates": "MSFT_MUSK000055001–MSFT_MUSK000055091 (approx.)", "pdf_url": "https://media.mts-in.com/PX-206.pdf", "body_markdown": "# PX 206 — OpenAI Global, LLC Amended and Restated LLC Agreement (January 23, 2023)\n\n> The 2023 successor to PX 201's 2019 LPA — the Microsoft \"$13B\" agreement on its face, restructured as an **LLC** after OpenAI, L.P. was converted to \"OpenAI OpCo, LLC\" and re-parented under a holding LLC (\"OpenAI Global, LLC\"). Schedule A discloses Microsoft's three closes — **$1B + $2B + $10B** Capital Commitments / **$20B + $12B + $60B = $92 Billion Microsoft Target Redemption** — alongside the $150B Atlas, LP / employee-vehicle target.\n\n## Document type\n**Corporate governance document — multi-doc compilation.** Bates-stamped Microsoft production (MSFT_MUSK series). Contains: (i) the **Amended and Restated Limited Liability Company Agreement of OpenAI Global, LLC** dated January 23, 2023 (Sections 1–10, ~75 numbered pages); (ii) the DocuSigned signature pages (Manager: OpenAI GP, L.L.C. by Sam Altman, CEO; Initial Member: OpenAI, L.L.C. by Sam Altman, CEO; Members: by OpenAI GP, L.L.C. as Attorney-in-Fact); (iii) **Schedule A — Member Information** with capital commitments and target redemption amounts; (iv) **Exhibit A — the OpenAI Charter** (re-attached); and (v) **Exhibit B — Reorganization Steps**.\n\n## Logistics\n- **Trial admission:** Day 7 (May 5, 2026). Used during the Robert Wu 30(b)(6) deposition designations on the 2023 Microsoft \"$13B\" deal — the contractual instantiation of the **$92B Microsoft target redemption** that anchors plaintiffs' \"$250B target-redemption stack\" arithmetic for Day 7.\n- **Box upload:** 2026-05-05 PT — Day 7 batch.\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~29.0 MB, 91 pages. **Note:** original PDF exceeded the 20 MB Read tool cap and was split into three parts in `Elon Musk, et al. v. Samuel Altman, et al - split/`.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `0206.pdf`.\n\n## Transcribed text\n\n> **AMENDED AND RESTATED LIMITED LIABILITY COMPANY AGREEMENT OF OPENAI GLOBAL, LLC**\n> **A DELAWARE LIMITED LIABILITY COMPANY**\n> **January 23, 2023**\n>\n> **IMPORTANT**\n>\n> *Investing in OpenAI Global, LLC is a high-risk investment*\n> *Investors could lose their capital contribution and not see any return*\n> *It would be wise to view any investment in OpenAI Global, LLC in the spirit of a donation, with the understanding that it may be difficult to know what role money will play in a post-AGI world*\n>\n> The Company exists to advance OpenAI Inc.'s mission of ensuring that safe artificial general intelligence is developed and benefits all of humanity. The Company's duty to this mission and the principles advanced in the OpenAI, Inc. Charter take precedence over any obligation to generate a profit. The Company may never make a profit, and the Company is under no obligation to do so. The Company is free to re-invest any or all of the Company's cash flow into research and development activities and/or related expenses without any obligation to the Members. See Section 6.4 for additional details.\n>\n> *[HIGHLY CONFIDENTIAL — MSFT_MUSK000055001]*\n\n> **Recitals (the conversion mechanics):**\n> \"WHEREAS, the Company was formed as a Delaware limited liability company pursuant to, and in accordance with the Act, by filing a certificate of formation of the Company ... with the Delaware Secretary of State on **December 28, 2022**;\n>\n> WHEREAS, upon its formation, the Company was governed by that certain Limited Liability Company Agreement of the Company entered into by **OpenAI, L.L.C.**, as its initial member (the 'Initial Member'), dated as of **January 23, 2023** (the 'Original LLC Agreement');\n>\n> WHEREAS, the Members (i) were previously members of **OpenAI OpCo, LLC (f/k/a OpenAI, L.P.)**, a Delaware limited liability company (that was converted from a limited partnership) ('OpenAI OpCo' ... ) and an indirect wholly-owned subsidiary of the Company, and (ii) received limited liability company interests in the Company by virtue of a **Reorganization** ... involving OpenAI OpCo and certain of its Affiliates ...\"\n>\n> The new top-of-stack entity is therefore **OpenAI Global, LLC**, with **OpenAI OpCo, LLC** (the converted-from-LP former OpenAI, L.P.) sitting underneath as a wholly-owned subsidiary, and **OpenAI GP, L.L.C.** acting as Manager.\n\n> **§ 2.3 Purpose and Scope:** \"Use the principles expounded in the Non-Profit Charter to advance the Non-Profit's mission of pursuing (i) the creation, development and refinement of a safe artificial general intelligence ... ('AGI'); and (ii) the dissemination and distribution of the benefits of such AGI among all of humanity (collectively, the 'Core Mission').\"\n\n> **§ 6.4 Subordination of Company and Member Interests to Core Mission; Modification of Fiduciary Duties** carries forward verbatim the all-caps mission-subordination and fiduciary-duty waiver from PX 201's § 6.4, with terminology updated for the LLC form (\"Manager,\" \"Members,\" \"Manager Related Persons\" instead of \"General Partner,\" \"Limited Partners,\" \"GP Related Persons\"). The substance — including the GP's authority to \"Retain any and all cash or property held by the Company as the Manager deems necessary to fund the acquisition of any business (including one or more large publicly-traded businesses) that the Manager reasonably and in good faith determines could achieve AGI (prior to the Declaration of Sufficient AGI), even if the manner of such achievement of AGI conflicts with the Core Mission\" — is preserved.\n\n> **§ 10.27 Reorganization** authorizes the Manager to take any further documentation step \"necessary or advisable in connection with the Reorganization\" — i.e., the LP-to-LLC conversion described in **Exhibit B (Reorganization Steps)**.\n\n> **Signature page (MSFT_MUSK000055082 / 206.082):**\n>\n> > **MANAGER: OPENAI GP, L.L.C.**, a Delaware limited liability company\n> > By: [DocuSigned — Sam Altman]\n> > Title: **Chief Executive Officer**\n> >\n> > **MEMBERS:** THE MEMBERS LISTED ON SCHEDULE A HERETO\n> > By: **OPENAI GP, L.L.C.**, a Delaware limited liability company\n> > Title: **Attorney-in-Fact**\n> >\n> > **INITIAL MEMBER: OPENAI, L.L.C.**, a Delaware limited liability company\n> > By: [DocuSigned — Sam Altman]\n> > Name: **Sam Altman**\n> > Title: **Chief Executive Officer**\n>\n> (Sam Altman therefore signs in three capacities — for the Manager, for the Initial Member, and as the attorney-in-fact for all of the Members under powers-of-attorney granted by Schedule A members.)\n\n> **SCHEDULE A — Member Information (As of January 23, 2023):**\n>\n> | Member | Capital Commitment | Capital Contributions | Target Redemption Amount |\n> |---|---|---|---|\n> | **Atlas, LP** (f/k/a OpenAI Holdings, L.P.) — the **employee/equity vehicle**; Jason Kwon, GC, c/o 575 Florida St., SF | $0 | $0 | **$150 Billion** (with $6,082,908,300 line) |\n> | **OpenAI, Inc.** (Non-Profit), 575 Florida St., SF; Sam Altman, CEO | **$60,829,083** | $60,829,083 | — |\n> | **Aphorism Foundation**, Menlo Park; Zack Herlick | $50,000,000 + $20,000,000 | matching | **$5B + $2B = $7B** |\n> | **Regents of the University of Michigan** | (redacted in extract) | — | — |\n> | **The Buchheit Revocable Trust** (Paul Buchheit) | $3,000,000 | $3,000,000 | $300,000,000 |\n> | **Khosla Ventures VI (AIV), L.P.** | $50,000,000 | $50,000,000 | $5,000,000,000 |\n> | **YC Holdings II AIV, LLC** (Y Combinator) | $10,000,000 | $10,000,000 | $1,000,000,000 |\n> | **Microsoft Corporation** (Second Close / Converted Member) — Attn: Keith Dolliver, VP & Deputy GC; and Michael Wetter, MD, Corporate Development | **$1,000,000,000** ** | $1,000,000,000 | **$20,000,000,000** |\n> | **Microsoft Corporation** (Third Close / Converted Member) | **$2,000,000,000** *** | $1,000,000,000¹ | **$12,000,000,000** |\n> | **Microsoft Corporation** (Fourth Close / Converted Member) | **$10,000,000,000** **** | $0 | **$60,000,000,000** |\n>\n> Footnotes: \"* Not including any adjustments pursuant to the definition of 'Target Redemption Amount.' ¹ Inclusive of funds in the Escrow Account which are part of the Third Closing Escrowed Amount. ** Second Closing Capital Commitment; subject to the conditions set forth in Section 3.2(b)(i) ... *** Third Closing Capital Commitment; subject to ... 3.2(b)(ii) ... **** Fourth Closing Capital Commitment; subject to ... 3.2(b)(iii) ...\"\n>\n> **Microsoft totals:** Capital Commitments **$13 billion** ($1B + $2B + $10B); Target Redemption Amounts **$92 billion** ($20B + $12B + $60B). Combined with the **$150B** Atlas/employee-vehicle target and the smaller LP targets (Aphorism $7B, Khosla $5B, YC $1B, Buchheit $0.3B, plus other LPs), the Schedule A redemption stack approaches the **$250B figure** Wu was walked through on Day 7.\n\n> **EXHIBIT A — OpenAI Charter** (re-attached from openai.com/charter — same text as PX 201 Exhibit A): mission to ensure AGI benefits all of humanity; Broadly Distributed Benefits (\"primary fiduciary duty is to humanity\"); Long-Term Safety (the \"stop-and-assist\" clause for value-aligned safety-conscious projects \"a better-than-even chance of success in the next two years\"); Technical Leadership; Cooperative Orientation.\n\n[Pages 30–75 of the LLC Agreement: Sections 6.5–10.28 — Expenses, Member/Manager Compensation, Records, Confidentiality (incl. Schedule A confidentiality), Disclosures, Holding/Feeder Vehicles, Transfers/Withdrawals/Removal, Dissolution and Liquidation, Liability and Indemnification, Meetings, Notices, Governing Law (Delaware), Investment Advisers Act / Exchange Act / Bad Actor / CFTC / Anti-Corruption / Whistleblower compliance, Section 10.18 (Akin Gump as counsel to the Manager and Company — explicitly not representing Members), and Section 10.27 (Reorganization). Standard LLC boilerplate adapting the 2019 LPA architecture to LLC form. Full text in source PDF: 0206.pdf, also split as 0206.part*.pdf.]\n\n[Pages 75–91: continued Schedule A entries for additional Members, addresses and side-letter references, **Exhibit B — Reorganization Steps** (the formal LP-to-LLC conversion sequence: OpenAI, L.P. → OpenAI OpCo, LLC; new holding company OpenAI Global, LLC formed Dec. 28, 2022; member interests rolled up; Atlas, LP renamed from OpenAI Holdings, L.P.). Full text in source PDF.]\n\n## Commentary\n\nPX 206 is the **2023 successor charter document** to PX 201. Together with PX 204 (the JDCA) and the Wu deposition designations, it lets plaintiffs build the Day 7 story arithmetic-by-arithmetic: the **$13 billion Microsoft \"investment\" buys $92 billion of Target Redemption Amount**, which the AGI definition in PX 204 § 4(i) makes an explicit financial trigger for the **\"Sufficient AGI\"** declaration. Three structural moves matter (see [[Day 7]] / [[Key Themes]] § \"Day 7 — Microsoft economics\"): (1) **The LP-to-LLC conversion** moves the for-profit out of the limited-partnership form (and out of the General Partner / Limited Partner fiduciary frame Brockman testified about) into a Delaware LLC where § 18-1101 of the Delaware LLC Act permits even broader contractual elimination of fiduciary duties — the all-caps § 6.4 carries forward verbatim. (2) **Sam Altman signs as Manager, Initial Member, and Attorney-in-Fact for all members** — the same triple-capacity dynamic plaintiffs flagged on the 2021 JDCA, now formalized in the holding-LLC structure. (3) **The \"Atlas, LP\" rename** of OpenAI Holdings, L.P. to \"Atlas, LP\" — the **$150B employee/equity vehicle target redemption** that anchors plaintiffs' \"$250B stack\" — is the OpCo-level commitment to the people building the system, alongside Microsoft's $92B and the legacy LP targets carried over from 2019. Akin Gump's role as counsel to the Manager and Company (and explicit non-counsel to the Members) is plaintiffs' answer to any defense argument that the documents were arms-length-negotiated for the LPs/Members. Cross-reference [[PX 201]] (the 2019 LPA this restates), [[PX 204]] (the JDCA whose \"Sufficient AGI\" trigger keys to these Schedule A numbers), and [[Brockman Journal]] / [[PX 157]] for the 2017 prophecy that the structure would matter more than any individual.\n\n---\n*See also:* [[Day 7|Day 7 digest]] · [[PX 201]] · [[PX 204]] · [[Robert Wu]] · [[Watershed MOU]] · [[Greg Brockman]] · [[Brockman Journal]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-207", "exhibit": "PX 207", "party": "Plaintiffs", "type": "Commercial agreement — Second Amended and Restated Joint Development and Collaboration Agreement (\"Argos II\"), with Exhibits A–I", "admitted_trial_day": "Day 7 (May 5, 2026)", "uploaded_box_pt": "2026-05-05 (Day 7 batch)", "uploader": "Someone", "pages": 70, "size_bytes": 26947584, "source_pdf": "PX-207.pdf", "split_parts": "0207.part1of3.pdf, 0207.part2of3.pdf, 0207.part3of3.pdf", "bates": "OPENAI_MUSK00010528–OPENAI_MUSK00010597 (approx.)", "pdf_url": "https://media.mts-in.com/PX-207.pdf", "body_markdown": "# PX 207 — \"Argos II\" Second Amended and Restated Joint Development and Collaboration Agreement (January 23, 2023)\n\n> The contemporaneous **2023 commercial counterpart** to PX 206 — the Microsoft / OpenAI Inc. / OpenAI OpCo, LLC master agreement (codenamed **\"Argos II\"**) that supersedes the 2021 JDCA in PX 204. Adds **20 / 80 revenue-share splits** on Net Revenue from OpenAI API and Microsoft New First Party Products, an **OpenAI-exclusive Azure Marketplace** commitment, a **Microsoft Capex Bucket** funding mechanism for the Phase 3/4 supercomputers, an **exclusive advertising provider** appointment, and the **\"Limited Exclusivity\"** clause: Microsoft will not pursue AGI other than jointly with OpenAI.\n\n## Document type\n**Commercial agreement — multi-party master contract.** Bates-stamped OpenAI production. Cover Pages with signature blocks (Microsoft signed by Satya Nadella, **Chairman and Chief Executive Officer**, dated 01/[x]/2026; OpenAI Inc. and OpenAI OpCo, LLC each signed by Sam Altman); General Terms (Sections 1–12); and Exhibits A (Specifications — Phase 2/3/4 supercomputers; Argos I and Argos II Dedicated Systems; Owl, Panda, Quail SPUs), B (Relationship Management / Governing Board / Trigger Date / Dispute Process), C (Trademarks), D (Deployment Safety Board), E (Support — incl. HPE Tier-1/2/3 schedule), F (Pricing and Payment — Capex Bucket mechanics), G (Transition), H (**New First Party Microsoft Product / Advertising Revenue** — 14 product-family carve-outs), and I (HPE Data Privacy and Security Agreement Schedule).\n\n## Logistics\n- **Trial admission:** Day 7 (May 5, 2026). Used during the Robert Wu 30(b)(6) deposition designations on Microsoft economics — the **commercial flip-side of PX 206's $13B / $92B target-redemption Schedule A**, with Argos II spelling out exactly what Microsoft is buying for that money.\n- **Box upload:** 2026-05-05 PT — Day 7 batch.\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~25.7 MB, 70 pages. **Note:** original PDF exceeded the 20 MB Read tool cap and was split into three parts in `Elon Musk, et al. v. Samuel Altman, et al - split/`.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `0207.pdf`.\n\n## Transcribed text\n\n> **SECOND AMENDED AND RESTATED JOINT DEVELOPMENT AND COLLABORATION AGREEMENT (Argos II)**\n>\n> This Agreement is among **Microsoft Corporation** ... (\"Microsoft\"); **OpenAI, Inc.** ... (\"OpenAI\"); and **OpenAI OpCo, LLC**, a Delaware limited liability company (formerly and successor to **OpenAI, L.P.** ('LP'), a Delaware limited partnership) ... (\"OpCo\"). This Agreement is effective on **January 23, 2023** (\"Effective Date\") and it amends, supersedes, and replaces in its entirety that certain **Amended and Restated Joint Development and Collaboration Agreement** entered into between the parties as of **March 5, 2021**, as amended (\"**Argos I Agreement**\") ...\n\n> **Contact details (Table 1):**\n> - OpenAI / OpCo: 3180 18th Street, Suite 100, San Francisco, CA 94110; **Attn: Sam Altman, CEO**; sama@openai.com; GC notice: **Jason Kwon** (jason@openai.com)\n> - Microsoft: One Microsoft Way, Redmond, WA 98052; **Attn: Kevin Scott, CTO**; GC notice: **VP, Deputy General Counsel, Frank Morrow**\n\n> **Signatures:**\n> - **Microsoft Corporation:** [DocuSigned] **Satya Nadella, Chairman and Chief Executive Officer** — Date of Execution: **01/[x]/2026** (interlineation in original suggests a 2026 amendment-and-restatement re-execution despite the January 23, 2023 effective date)\n> - **OpenAI, Inc.:** [Sam Altman]\n> - **OpenAI OpCo, LLC:** [Sam Altman]\n\n> **Background recitals:**\n> \"The parties subsequently entered into the Argos I Agreement to accommodate certain changes in their business relationship as originally set forth in the Original Agreement ... Contemporaneously with the Argos I Agreement, the parties entered into that certain **Second Amended and Restated** [Limited Liability Company Agreement] ... The parties now wish to modify their existing agreement to accommodate certain changes in their business relationship as modified and set forth in the Argos I Agreement, **providing for additional financial commitment** [from Microsoft] ...\"\n\n> **§ 1 Definitions (selected):**\n> - **\"Argos I Dedicated System\"** means, collectively, (i) the Supercomputing Processing Units known to the parties as of the Effective Date as **Owl, Panda, or Quail**; and (ii) the Phase 2 Supercomputer.\n> - **\"Argos II Dedicated System\"** means, collectively, (i) the Phase 3 Supercomputer, (ii) the Phase 4 Supercomputer, and (iii) any related supercomputer system purchased by OpenAI under this Agreement.\n> - **\"Sufficient AGI\"** has the meaning set forth in Section 4(k)(i); **\"Future Developments\"** in Section 4(k)(ii); **\"Trigger Date\"** in Exhibit B; **\"Third Investment Agreement\"** referenced in Background; **\"Capex Bucket\"** referenced in §§ 4 / 5 and Exhibit F.\n\n> **§ 3 IP Licenses (Argos II structure):**\n> § 3(c) grants Microsoft \"a perpetual, irrevocable, worldwide, **exclusive (solely during the Term, and except for the limited rights retained by OpenAI)**, royalty-free, and fully paid license under all Licensed IP and for all purposes.\" (i.e., Argos II is the exclusivity escalation from the 2021 JDCA's narrower exclusivity windows.)\n>\n> § 3(f) **Effect of Termination, Expiration, or Achievement of Sufficient AGI:** \"When this Agreement expires or is terminated, or **on the achievement of Sufficient AGI**, or upon a default being triggered under Section 11(b) of the Third Investment Agreement, the license granted in Section 3(c) **will become non-exclusive** ...\"\n\n> **§ 4(j) Revenue-Sharing (the \"20/80\" splits):**\n> § 4(j)(iii)(A) — **Net Revenue generated by OpCo and OpenAI:** \"OpCo and OpenAI will retain 80%, and Microsoft and its Affiliates will receive **20% of Net Revenue generated by OpCo and OpenAI during the Term through the Commercial Use of any Technology embodying any Licensed IP** as permitted in this Agreement, whether such Commercial Use is through the OpenAI API or otherwise.\"\n>\n> § 4(j)(iii)(B) — **Net Revenue generated by Microsoft:** \"Microsoft and its Affiliates will retain 80%, and OpCo and OpenAI will receive **20% of Net Revenue generated by Microsoft or its Affiliates during the Term through third-party Commercial Use of the Microsoft API and the Commercial Use of any New First Party Microsoft Products**.\"\n\n> § 4(j)(iv) — **Transition of OpenAI IP to Azure Marketplace:** \"No later than **six months after the Effective Date**, OpCo and OpenAI will transition the OpenAI API to be available solely through the Azure Marketplace. ... After completion of the transition, and for the remainder of the Term, **OpCo and OpenAI will exclusively commercialize the OpenAI API on the Azure Marketplace** (and any successor versions to such Azure Marketplace).\"\n\n> § 4(j)(v) — **Exclusive Advertising Provider:** \"The parties will, within 45 days after the Effective Date, negotiate and execute a separate contract under which OpenAI and OpCo will grant to Microsoft the **exclusive right to solicit, deliver, and display advertising in all products, services, and Technologies in or through which OpenAI, OpCo, or any of their respective Affiliates includes advertising**, subject to OpenAI's right to directly solicit such advertising without the assistance of any third party.\"\n\n> **§ 4(k) Artificial General Intelligence (the \"Sufficient AGI\" provision — Argos II form):**\n>\n> > **\"(i) Achievement.** OpCo and OpenAI may during the Term, in good faith, declare certain Models developed by OpCo or OpenAI to be '**Sufficient AGI**,' meaning that such Models: (A) are within the scope of the definition of AGI; and (B) are clearly capable of generating the **Target Redemption Amount (defined in the Third Investment Agreement) of return on investment capital for each of OpenAI's members (including Microsoft)**, which such determination will be made pursuant to standards that **Microsoft has reviewed and approved (such approval not to be unreasonably withheld, conditioned, or delayed)**. On achievement of Sufficient AGI as provided in this Section 4(k)(i), OpCo and OpenAI may collaborate with Microsoft to monetize Sufficient AGI **until OpCo and OpenAI generate and pay over to each member the full Target Redemption Amount** payable to each such member.\n> >\n> > **(ii) Future Developments.** After OpCo and OpenAI pay over to each such member the full applicable Target Redemption Amount in accordance with Section 4(k)(i): (A) OpCo, OpenAI, and Microsoft will collaborate in good faith to make the benefits of artificial general intelligence available to all humanity ('**AGI Distribution**'); and (B) **Microsoft will have the right of first refusal to be OpCo's and OpenAI's commercialization partner for AGI** if OpCo or OpenAI chooses to have a commercialization partner ...\n> >\n> > **(iii) Limited Exclusivity.** Throughout the Term: (A) **OpCo and OpenAI will use Microsoft as the preferred compute partner for AGI**; and (B) **Microsoft will not pursue AGI** (including investments in or collaboration with third parties pursuing AGI, in each case solely to the extent such investments or collaborations are directed to AGI) **other than jointly with OpCo or OpenAI**.\"\n>\n> § 4(l) Tax Treatment: \"the parties and their respective Affiliates agree to treat **Microsoft's funding of the Capex Bucket** in a manner consistent with the treatment provided in the **Third Investment Agreement**.\"\n\n> **§ 5(a) Microsoft Cloud and Edge Services:** \"OpCo and OpenAI will, throughout the Term, **use solely Microsoft's Azure services** (and other relevant Microsoft cloud, edge, and supercomputing services) for **all artificial intelligence training, Inference, deployment, and commercial workloads**. No purchases of such Azure services will be deducted from the Capex Bucket. ... For the avoidance of doubt, OpenAI is under no obligation to refrain from using any third party service for purposes other than artificial intelligence training, Inference, deployment, and commercial workloads (e.g., HR, financial systems, email, business communications, etc.).\"\n\n> **§ 5(b) Governance:** \"The parties' relationship under this Agreement will be managed by the **Governing Board**, which will be comprised of personnel from each party as specified in Exhibit B ... For purposes of determining composition of the Governing Board, OpCo and OpenAI will be jointly represented so that the **aggregate number of representatives of these two parties on the Governing Board will be the same as the number of Microsoft representatives**.\"\n\n> § 11(b)(iii) **Effect of Termination — AGI:** \"Microsoft's rights and privileges and OpenAI's and OpCo's obligations under Section 4(k) with regard to AGI and Sufficient AGI **will be deemed removed in their entirety on a prospective basis and not replaced.**\"\n\n> **Exhibit F — Capex Bucket mechanics (selected):**\n> \"All Capex Bucket deductions made pursuant to this Agreement will be made quarterly, in arrears.\" Argos I Dedicated System pricing (excluding Owl) and Argos II Dedicated System purchases are paid by deduction from the Capex Bucket; **Microsoft has no obligation to purchase a new Argos II Dedicated System if at the time the cumulative deductions plus the proposed purchase exceed the remaining amounts of the Capex Bucket.**\n\n> **Exhibit H — New First Party Microsoft Product / Advertising Revenue (Section 1):** \"Microsoft will share revenue from New First Party Microsoft Products in accordance with Section 4(j)(iii)(B).\" A **\"New First Party Microsoft Product\"** is defined as a Microsoft product or service for broad commercialization that is not (a) a \"Current Product\" — defined by 14 enumerated families: **Advertising Products, Business Application Products, Cloud Platform Products, Gaming Products, Healthcare Products, Information Worker Productivity Products, Operating System Products, Personal Computing Devices, Professional Social Network Platforms, Search and Information Products, Security Products and Functionality, Software Development Products, Unified Communications Products, and Virtual or Augmented Reality Functionality** — or (b) a successor to, (c) in the same family as, or (d) substantially derived from one or more Current Products. The In-Scope Products for advertising revenue sharing are **Microsoft Bing, MSN, and Microsoft Start** (and successors).\n\n[Pages 8–35: §§ 5 (Other Rights and Obligations — Governing Board, Designated/Trusted Engineers, DSB, Joint Marketing), 6 (Publicity / Joint Marketing), 7 (Confidentiality / NDA), 8 (Reps & Warranties), 9 (Compliance / China carve-outs / export), 10 (Defense of Third-Party Claims), 11 (Term — initial term plus renewal; insolvency), 12 (Notices / Governing Law / Interpretation). Exhibits A (Phase 2/3/4 supercomputer specs and Argos I/II Dedicated Systems; Owl, Panda, Quail SPUs), B (Governing Board composition, Trigger Date, Dispute Process). Full text in source PDF: 0207.pdf, also split as 0207.part*.pdf.]\n\n[Pages 36–70: Exhibits C (Trademarks), D (Deployment Safety Board with Safety-Critical Parameters Threshold and Safety-Critical Compute Threshold carried over from Argos I), E (Support — Tier 1/2/3 obligations, HPE handoff for Professional Services Data tickets), F (Capex Bucket Pricing & Payment), G (Transition — including treatment of Argos I exclusivities and the **GitHub/Codex six-year exclusivity** carried forward), H (New First Party Microsoft Product/Advertising Revenue — full 14-family definition; advertising-revenue-growth calculation and base-period mechanics), and I (**HPE Data Privacy and Security Agreement Schedule** — physical security standards, processor obligations, GDPR/Schedule of Sub-Processors). Full text in source PDF.]\n\n## Commentary\n\nPX 207 (\"**Argos II**\") is the **commercial settlement of the Microsoft–OpenAI relationship** that plaintiffs pair with PX 206 (the LLC Agreement) and PX 204 (the Argos I JDCA) on Day 7 to make the Microsoft-economics case (see [[Day 7]] / [[Key Themes]] § \"Day 7 — Microsoft economics\"). Three Argos II provisions do the heaviest plaintiff work: (1) **§ 4(k) \"Sufficient AGI\" (carried forward from Argos I and amplified)** — the AGI definition is again a financial trigger keyed to **paying out the Target Redemption Amounts under the Third Investment Agreement** (which the Schedule A in PX 206 totals at ~$250B), with Microsoft holding the approval pen (\"standards that Microsoft has reviewed and approved, such approval not to be unreasonably withheld, conditioned, or delayed\"); (2) **the new § 4(k)(iii) \"Limited Exclusivity\"** is a striking commitment for a defendant whose narrative is independence — Microsoft contractually promised to **not pursue AGI other than jointly with OpenAI**, and OpenAI contractually promised to use Microsoft as the preferred compute partner for AGI, both throughout the Term; and (3) **the 20/80 revenue splits, the Azure-Marketplace exclusivity for the OpenAI API, the all-AI-workloads-on-Azure-only commitment in § 5(a), and Microsoft's exclusive advertising provider appointment**, taken together, mean Microsoft has bought not only an effective monopoly on OpenAI commercialization for the Term but also a **20% revenue tap on every dollar OpenAI earns from any Licensed-IP product** in addition to keeping 80% of its own AI revenue. The signature page is the eyebrow-raiser: Nadella's date of execution reads **\"01/[x]/2026\"** (in handwritten interlineation over the printed effective date), suggesting either a re-execution or amendment-and-restatement that landed on the docket alongside the public **Watershed MOU** discussions of Day 7. **Sam Altman again signs in two capacities** (OpenAI Inc. + OpCo). Cross-reference [[PX 206]] (the LLC Schedule A this contract economically funds), [[PX 204]] (the Argos I JDCA this supersedes), [[PX 201]] (the original 2019 LPA), [[Watershed MOU]], and [[Brockman Journal]] for the long Brockman/Cerebras compute-partner concerns from the 2017 control fight.\n\n---\n*See also:* [[Day 7|Day 7 digest]] · [[PX 201]] · [[PX 204]] · [[PX 206]] · [[Robert Wu]] · [[Watershed MOU]] · [[Greg Brockman]] · [[Brockman Journal]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-211", "exhibit": "PX 211", "party": "Plaintiffs", "type": "Term sheet", "admitted_trial_day": "Day 7 (May 5, 2026)", "uploaded_box_pt": "2026-05-05T16:39:48", "uploader": "Someone", "pages": 14, "size_bytes": 2222182, "source_pdf": "PX-211.pdf", "pdf_url": "https://media.mts-in.com/PX-211.pdf", "body_markdown": "# PX 211 — September 11, 2025 OpenAI–Microsoft \"Watershed\" Memorandum of Understanding (PBC Recapitalization)\n\n> Executed but non-binding (except governing law) MOU between OpenAI, Inc. and Microsoft Corporation setting out the terms of the proposed recapitalization of the OpenAI for-profit enterprise into a Delaware Public Benefit Corporation — the document Robert Wu authenticated on Day 7 and that plaintiffs use to argue the 2018 Charter's AGI carve-out is dissolving.\n\n## Document type\n**Term sheet (Memorandum of Understanding) — Confidential / Execution Version.** Multi-section term sheet with capitalization annex (Annex A) and signature pages. Executed by Sam Altman as CEO of OpenAI, Inc. and by Amy Hood (Sep 11, 2025 12:28:13 PDT) as EVP and CFO of Microsoft Corporation. Bates OPENAI_MUSK00037469 through OPENAI_MUSK00037482.\n\n## Logistics\n- **Trial admission:** Day 7 (May 5, 2026) — Brockman defense direct + Mr. Molo recross + DX 1252; Wu deposition.\n- **Box upload:** 2026-05-05 16:39:48 PT — Day 7 late-afternoon upload, contemporaneous with the Wu video deposition.\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~2.12 MB, 14 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `211.pdf`.\n\n## Transcribed text\n\n**[Cover / Header]**\n\n> Confidential — Execution Version\n>\n> **MEMORANDUM OF UNDERSTANDING**\n> **Proposed Recapitalization of the OpenAI For-Profit Enterprise**\n> **September 11, 2025**\n>\n> This Memorandum of Understanding summarizes certain terms relating to a potential recapitalization of the OpenAI for-profit enterprise. This Memorandum of Understanding does not contain all the terms, conditions and other provisions of the proposed transaction. With the exception of the Section entitled \"Governing Law\" (which shall be binding), this Memorandum of Understanding is not intended to constitute, and does not constitute, a legally binding agreement or an offer to consummate the contemplated transaction. This Memorandum of Understanding and the information contained herein are strictly confidential.\n\n**[Transaction Principles]**\n\n> The guiding principle for OpenAI, Inc. (the **NFP**) is and continues to be the charitable mission, which is to develop artificial general intelligence for the benefit of all humanity. The expansion of the OpenAI for-profit enterprise and evolution of the AI competitive landscape, among many other factors, have presented unique challenges for the NFP and its ability to further the mission. It is critical to the success of the mission that the for-profit enterprise remain a technical leader in developing AGI, as well as advancing its use by individuals and enterprises. **In the view of the NFP Board, the current structure hinders its ability to fulfill the mission, and a change in the structure of the group is essential.**\n>\n> To address these concerns and promote the sustainability of the OAI for-profit enterprise for the long term, the NFP Board is considering a potential recapitalization of the for-profit enterprise. **In the recapitalization, the OAI for-profit enterprise will be held under a newly formed Delaware public benefit corporation (PBC), and the PBC will continue to be controlled by the NFP.** The NFP will have control of the PBC through its control of the election of PBC directors as well as certain pre-approval rights with respect to PBC matters (e.g., changes to the PBC's mission, a sale of control of the PBC and any sale of material assets by the PBC).\n>\n> The mission of the PBC will be identical to the mission of the NFP.\n>\n> The transaction also contemplates significant control by the NFP over safety and security matters of the OAI for-profit enterprise, as well as full decision making alignment with the mission with respect to those matters.\n\n**[Transaction Structure]**\n\n> The OpenAI for-profit enterprise will be reorganized under a PBC. The recapitalization will be implemented through a series of simultaneous internal mergers and other transactions described below under the heading \"Recapitalization Steps\" resulting in: (a) the NFP, (b) the First Closing LPs (**FCLPs**), (c) Aestas, LLC (i.e., the vehicle through which employees, ex-employees, tender investors and ex-Rockset shareholders hold their interests indirectly through Aestas Management Company, LLC) (**Aestas**), (d) investors in OpenAI's October 2024 financing round (**Oct. 2024 Investors**), (e) investors in OpenAI's 2025 financing round being led by SoftBank Group Corp. (**Softbank**, and such investors collectively, **Sakura Investors**), and (f) any equity holders whose interest arises from selling a business (e.g., Project Icon, Project Stallion) to the OpenAI for-profit enterprise (**M&A Sellers**), with each of (a)…(f) holding equity of the PBC directly. **Microsoft Corporation (Microsoft) would continue to own interests in OpenAI Global, LLC (OAI Global), at a level below the PBC, which interests will be exchangeable at Microsoft's option for PBC equity, subject to obtaining any applicable regulatory approvals.**\n>\n> At the time of the NFP Board's approval of the recapitalization terms, Microsoft, the NFP and OAI Global will execute a framework agreement (the **Framework Agreement**) pursuant to which, among other things, **Microsoft will deliver its irrevocable consent to the recapitalization** of the OpenAI for-profit enterprise on terms consistent with this MOU. The recapitalization will be consummated immediately thereafter.\n\n**[Required Approvals]**\n\n> *Consents.* The recapitalization will not require any consents, other than from the NFP Board and Microsoft.\n>\n> *Notices.* At least 14 days' notice is required to be provided to the Delaware AG prior to the NFP Board meeting to finally approve the recapitalization. Practically, we expect to engage proactively with both the Delaware and California AGs well in advance of a decision; engagement is currently ongoing.\n>\n> SoftBank is entitled to 20 business days' prior notice (no consent) of completion of the recapitalization. Certain Oct. 2024 Investors who invested through blocker entities are also entitled to 10 business days' prior notice (no consent).\n\n**[Capital Structure — Capital Stock]**\n\n> 1) **Class N Common Stock.** 1 million shares issued to the NFP only, in consideration of its contribution of OAI GP to the PBC. **Non-transferable**; holds $1 per share of par value but otherwise has no economic value. Class N Common Stock holds 100% of the vote in director elections when no Class B Common Stock is outstanding. When there are shares of Class B Common Stock outstanding, the Class N Common Stock will hold a 2/3rds supermajority of the voting power in director elections. Also has specified pre-approval rights described herein, in each case solely to the extent held by the NFP.\n> 2) **Class A Common Stock.** Initially issued to NFP (with respect to its residual interest), Aestas and M&A Sellers. Entitled to vote on all matters, except for director elections.\n> 3) **Class B Common Stock.** Initially unissued. Held in reserve for acquisitions or similar strategic transactions...\n> 4) **Series A-1 Preferred Stock.** Initially held by FCLPs and the NFP (with respect to its FCLP interest). Convertible preferred stock with 1x preference of invested capital (measured as the FCLPs' and NFP's capital contributions to OAI Global of approximately **$193 million**).\n> 5) **Series A-2 Preferred Stock.** Initially held by Oct. 2024 Investors. Convertible preferred stock with 1x preference (in an amount equal to their investment amount of approximately **$6.7 billion** (and up to $7.7 billion if Thrive exercises option to invest in second tranche)).\n> 6) **Series A-3 Preferred Stock.** Initially held by SoftBank and other Sakura Investors. Convertible preferred stock with 1x preference (in an amount equal to their investment amount of **up to $41 billion**).\n> 7) **Series B Preferred Stock.** Initially unissued. **Microsoft's OAI Global units will be exchangeable for Series B Preferred Stock** (or, if there has been a mandatory conversion event, Class A Common Stock) with 1x preference (in an amount equal to Microsoft's invested opex [and capex] of approximately **$[13.0] billion**) (see \"Up-C Terms,\" including potential forfeiture in connection with Default Remedies).\n>\n> All classes of Preferred Stock are convertible into Class A Common Stock... All Preferred Stock will automatically convert into Class A Common Stock **upon an IPO or other go-public transaction**.\n\n**[Retained NFP Approval Rights]**\n\n> So long as the NFP continues to hold Class N Common Stock, the NFP will have a pre-approval right to be further described in definitive agreements with respect to: (1) any change to the PBC's mission; and (2) a sale of control of the PBC or a sale of material assets by the PBC...\n>\n> In addition, by virtue of being the sole holder of Class N Common Stock, the NFP will always control the right to appoint PBC directors and, subject to limited exceptions described below, remove directors.\n\n**[Up-C Terms — Microsoft]**\n\n> Subject to Microsoft's ongoing tax and regulatory review, **Microsoft would continue to hold a direct investment in OAI Global, which will be exchangeable for Series B Preferred Stock at Microsoft's option** at an exchange ratio designed to preserve Microsoft's as-converted ownership percentage. Except as may be otherwise agreed by the parties in the definitive agreements, upon Microsoft exercising its option to exchange, OAI Global would have the option to either redeem Microsoft's interest for shares of Series B Preferred Stock or an amount of cash equal to the value of such stock... (\"**Redemption**\")...\n>\n> OAI Global would continue to be treated as a partnership for federal income tax purposes and be owned jointly by Microsoft and the PBC...\n>\n> In order to maintain the fixed exchange ratio, the PBC will generally be required under the terms of an amended LLC agreement of OAI Global (the **New Global LLCA**) to operate all of its businesses through OAI Global and its subsidiaries (i.e., all of the PBC's assets and liabilities would be owned directly or indirectly by OAI Global) and also to issue or cancel/repurchase interests in OAI Global, as applicable, to mirror future changes in the PBC's capitalization.\n>\n> The Default Remedies (as defined in the current OAI Global LLC agreement) will continue to apply with respect to Microsoft's failure to fund its remaining capital contributions...\n\n**[Aestas Wind-down]**\n\n> In connection with the recapitalization, Aestas and Aestas ManagementCo will each adopt a plan of liquidation that will span approximately two years, subject to potential additional buffer for new hire and other grants promised but unissued as of the closing of the recapitalization (the \"**Liquidation Period**\"). The plan of liquidation will provide that, in general, awards held by Aestas Holders will convert to PBC common shares (or awards in respect of PBC common shares) either at the time of recapitalization, during the Liquidation Period or at the end of the Liquidation Period, as applicable.\n\n**[Voting]**\n\n> Only the NFP will be entitled to vote on PBC director elections (by virtue of being the sole holder of Class N Common Stock) and, subject to limited exceptions described below, to remove directors, for so long as there are no shares of Class B Common Stock outstanding. To the extent there are shares of Class B Common Stock outstanding, the ratio of Class N to Class B Common Stock will give the NFP a 2/3rds supermajority vote for director elections...\n>\n> Otherwise, the PBC's Charter will provide that holders of capital stock and Microsoft will be entitled to vote on all other matters on which stockholders are entitled to vote under applicable law (e.g., mergers, charter amendments).\n\n**[Board Duties]**\n\n> The NFP directors, in their capacity as such, remain duty bound to the NFP's mission of ensuring that AGI benefits all humanity.\n>\n> As required by Delaware law under the public benefit corporation statute, **PBC directors, in their capacity as such, will have a duty to take into account (a) the mission, (b) the best interests of those materially affected by the PBC's conduct, and (c) the stockholders' financial interests.** Members of the SSC will solely consider the mission and not the interests of stockholders and other stakeholders of the PBC in serving on the SSC.\n>\n> The PBC's Charter will specify a mission identical to the NFP's charitable purpose set forth in its current charter. As described above, any change to the PBC's mission will require the NFP's consent.\n\n**[Board Composition]**\n\n> *Initial PBC Board.* All NFP Board members will also be PBC Board members immediately following the recapitalization.\n>\n> *Ongoing PBC Board.* Nominees for the PBC Board will be selected exclusively by the NFP. The Corporate Governance Guidelines, amendments to which will be subject to the NFP's consent, will require that at all times at least a 2/3rds supermajority of NFP directors serve as a 2/3rds supermajority on the PBC Board...\n\n**[PBC Board Committees]**\n\n> 1) **Audit and Risk Committee.** To consist of at least three directors (all independent).\n> 2) **Comp Committee.** To consist of at least two directors (all independent).\n> 3) **Nom/Gov Committee.** To consist of at least three directors (all independent).\n> 4) **Safety & Security Committee.** To consist of at least three directors (all independent and all of whom also serve as NFP directors). **SSC decisions will be driven solely by the best interests of the charitable mission.**\n\n**[PBC Board Decisionmaking — supermajority items]**\n\n> A 2/3rds supermajority of directors then eligible to vote on a matter would be required to:\n> 1) Amend the PBC's organizational documents in any material respect (excluding ministerial amendments and fundraising related amendments)\n> 2) Remove the CEO\n\n**[Investor Rights — Microsoft side letter cancellation]**\n\n> For the avoidance of doubt, other than information rights, **the rights in the Side Letter dated January 23, 2023 between OAI Global and Microsoft (including without limitation the right of first refusal on fundraisings and right of first notification for tender offers) shall no longer be in effect after the recapitalization.**\n\n**[Non-Profit collaboration]**\n\n> Support from, and collaboration with, the PBC following the recapitalization will be critical to the NFP's viability to achieve its mission... The NFP and PBC will enter into a collaboration agreement, pursuant to which **the NFP will have access to PBC intellectual property and personnel to support its charitable purpose which will be provided to the NFP without charge**, as well as access to liquidity, further details of which will be set forth in definitive agreements.\n>\n> It is critical to the viability of the recapitalization that the NFP remain a \"**public charity**\" for U.S. federal income tax purposes...\n\n**[Governing Law]**\n\n> This Memorandum of Understanding shall be governed by the laws of the State of Delaware.\n>\n> Other than this Section entitled \"Governing Law\", this Memorandum of Understanding is not intended to and does not create any legally binding obligation of any nature on any party.\n\n**[Signatures]**\n\n> **OpenAI, Inc.** — By: /s/ Sam Altman — Title: Chief Executive Officer\n> **Microsoft Corporation** — By: /s/ Amy Hood (Sep 11, 2025 12:28:13 PDT) — Title: EVP and CFO\n\n**[Annex A — Pro Forma Cap Table — Waterfall]**\n\n> The below table shows pro forma ownership percentages before giving effect to (i) dilution from the convertible securities held by the Sakura Investors and Oct. 2024 Investors and (ii) adjustments for securities issued to M&A Sellers. In addition, **the NFP will receive warrants for shares of Class A Common Stock, with the number of shares, strike price, tenor and other terms to be mutually agreed and resulting in a present value at closing of $22.4 billion, calculated based on a Black-Scholes methodology using 50% volatility (the \"NFP Warrants\").** The NFP Warrants will dilute all PBC shareholders on a pro rata (as-converted) basis. The Aestas percentage below will be distributed against 280 million Aestas units.\n>\n> | Holder | % |\n> |---|---|\n> | Aestas | **33.0%** |\n> | Microsoft | **32.5%** |\n> | FCLPs (excluding the NFP's FCLP interest) | **3.0%** |\n> | NFP (including its FCLP interest) | **31.5%** |\n\n## Commentary\n\nPX 211 is the [[Key Themes|\"Watershed MOU\"]] referenced throughout Robert Wu's Day 7 deposition — the term sheet under which the OpenAI for-profit enterprise reorganizes into a Delaware Public Benefit Corporation, with the NFP holding a non-economic Class N supermajority for director elections and Microsoft's existing OAI Global stake exchangeable for new Series B Preferred Stock with a **~$13B 1x preference**. **Plaintiffs' theory of breach** gains a 2026-vintage supporting fact here: Wu testified that \"**Under the Watershed construct, Microsoft would receive — once there is a determination that there is AGI through this third-party panel — certain access to certain product IP**\" — meaning the 2018 Charter's \"**AGI is excluded from any commercial license**\" carveout is being dissolved precisely as plaintiffs allege. The Sakura/SoftBank A-3 preference of **up to $41 billion** plus the Oct. 2024 Investors' **$6.7B (up to $7.7B)** plus Microsoft's **$13B** — stacked on top of all the other LP preferences Wu testified added up to **~$250 billion** in target redemptions before the nonprofit residual gets a dollar — is the structural picture plaintiffs put before the jury. The **$22.4B Black-Scholes-valued NFP Warrants** in Annex A are the consideration paid to the nonprofit, against the **33.0%/32.5%/31.5%** Aestas/Microsoft/NFP cap table — concrete numbers for the [[Day 7|Day 7]] \"is the nonprofit getting fair value?\" question. See also [[Key Themes]] for the broader Microsoft-economics arc and [[Day 7|Day 7 digest]] for Wu's authentication and live testimony.\n\n---\n*See also:* [[Day 7|Day 7 digest]] · [[Robert Wu Deposition]] · [[Microsoft Thread]] · [[Key Themes]] · [[Case Overview]]\n"}
{"exhibit_id": "PX-212", "exhibit": "PX 212", "party": "Plaintiffs", "type": "Term sheet", "admitted_trial_day": "Day 7 (May 5, 2026)", "uploaded_box_pt": "2026-05-05T16:39:49", "uploader": "Someone", "pages": 20, "size_bytes": 10548736, "source_pdf": "PX-212.pdf", "pdf_url": "https://media.mts-in.com/PX-212.pdf", "body_markdown": "# PX 212 — \"Argos 3\" Microsoft / OpenAI Watershed term sheet\n\n> The non-binding \"Argos 3\" summary of key terms between OpenAI OpCo, OpenAI Inc., and Microsoft — the **Watershed MOU** Robert Wu authenticated on Day 7 — restructuring the AGI carveout, the IP-transfer regime, the Disclosure Team, the OE program, the recapitalization split (Microsoft preferred equity at **32.5%** fully diluted), and the IP Extension Period through December 31, 2032. Signed by **Sam Altman** (OpenAI OpCo and OpenAI Inc.) and **Amy Hood** (Microsoft).\n\n## Document type\n**Term sheet, 20 pages including signature pages, Exhibit A (Model BoM), and Exhibit B (SLA Table).** Bates OPENAI_MUSK00037449–00037468. Stamped HIGHLY CONFIDENTIAL / MICROSOFT & OPENAI CONFIDENTIAL on every page. Header: \"**ARGOS 3 — INTENDED SOLELY AS A SUMMARY OF KEY TERMS AND DOES NOT CONSTITUTE A BINDING AGREEMENT**.\" Sections: AGI; Licensed IP; IP Transfer (largely redacted on the public copy); OE Program / Support Team / Safety / OSS / Government; Other Terms (revenue share, recapitalization, Extension IP, definitive agreements). DocuSigned by Altman; Hood signed Sept 11, 2025.\n\n## Logistics\n- **Trial admission:** Day 7 (May 5, 2026) — Brockman defense direct + Mr. Molo recross + DX 1252; Wu deposition. The exhibit is the document Wu referred to throughout his 30(b)(6) testimony as the \"Watershed MOU\" (see [[Key Themes]] §\"The Watershed MOU — and the AGI carveout dissolving\").\n- **Box upload:** 2026-05-05 16:39:49 PT — late-afternoon batch on Day 7.\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~10.5 MB, 20 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `0212.pdf`.\n\n## Transcribed text\n\n> **ARGOS 3**\n> *INTENDED SOLELY AS A SUMMARY OF KEY TERMS AND DOES NOT CONSTITUTE A BINDING AGREEMENT*\n>\n> **AGI**\n>\n> **1) AGI Relevant Research**\n> a) OAI may withhold any AGI Relevant Research at any time during the term of the agreement subject to Sections 1(c) and 3(a) below. OAI will provide notice to Microsoft immediately upon its decision to withhold any AGI Relevant Research from transfer to Microsoft.\n> b) **\"AGI Relevant Research\"** means confidential, non-public research that, in OAI's good faith judgment, is reasonably expected to materially advance the state of the art in a manner that enables a model to satisfy the definition of AGI, and does not include research that contributes in an incidental or ancillary way, without materially influencing the state of the art.\n> c) Except as set forth in Section 1(d) below, OAI may not commercialize a system or model resulting from AGI Relevant Research without Microsoft's consent.\n>   i) For clarity, OAI may disclose a system or model resulting from AGI Relevant Research to third parties under NDA for red teaming and other non-commercial testing and evaluation.\n> d) If OAI submits a system resulting from AGI Relevant Research to the AGI expert panel (explained below), then OAI may commercialize that system, and Microsoft will receive the AGI Model BOM (to be defined) for that system in accordance with the IP transfer process.\n> e) In good faith, if OAI determines that any withheld AGI Relevant Research is not likely to result in system or model that meets the definition of AGI, or decides that it will not submit a system or model resulting from that AGI Relevant Research to the panel, it will transfer the AGI Relevant Research to Microsoft promptly.\n>\n> **2) AGI Expert Panel**\n> a) **Panel Composition and Selection Process.** During the Declaration Window, OAI may declare a system to be AGI and initiate the process of appointing a panel of experts.\n>   i) Panelist Qualifications:\n>     (1) **AI Expert** — must have prior work experience in AI technical/research role at major AI lab other than OAI or Microsoft (i.e., not primarily academic experience)\n>     (2) **Economist Expert** — must have a PhD in economics and have published about the economics of AI in a leading peer-reviewed journal\n>     (3) **Legal Expert** — former federal judge with experience in contract interpretation, technology and IP.\n>   ii) Panel Selection Process. The panel will be comprised of five individuals: (i) two Economics Experts, (ii) two AI Experts, and (iii) one Legal Expert. To choose the panel, each party will draft and submit to the other side, within 10 days of OAI's initiation of the Panel Selection Process, a list of five candidates for each member of the panel. Any individual common to both parties' lists shall be automatically selected. … \"Cause\" shall be defined to include a lack of independence or the appearance thereof. … The Panel Selection Process shall take no longer than thirty days.\n>   iii) Panelist Communications. … Aside from the diligence questionnaire, a party may not discuss this matter with a potential panelist. … Once the panel is appointed, neither party may communicate with any panel member (other than jointly) until the conclusion of the engagement.\n>   iv) Replacement of Panelists.\n> b) Role and process:\n>   i) The panel will assess and decide whether the declared system meets the definition of AGI as defined in the JDCA. … a majority of the panel members, with at least one from each category, must agree that the system meets the definition for it to be deemed AGI under the JDCA.\n>   ii) The panel will render its decision within 30 days of the engagement of all panel members. It may take up to an additional 60 days … but it must render its decision no later than 90 days after engagement.\n>   iii) **Independent Research and Communications.** OAI will submit materials to the panel describing the basis for its declaration … (though OAI may withhold any AGI Relevant Research, if requested by the panel, from Microsoft).\n>   iv) **Confidentiality.** The entire panel process … shall remain confidential and any such obligation may be waived only with the express agreement of both parties.\n>   v) **Public Disclosure.** Microsoft and OAI will cooperate and coordinate on proactive public disclosure or response to unauthorized disclosure.\n>   vi) **Ongoing Obligations of Independence.**\n>   vii) **Continuity of the Panel.**\n>   viii) **Costs.** The Parties shall bear their own fees and expenses … The Parties will otherwise equally share costs equally, including the payment of the members of the Panel.\n>\n> **3) Effects of the Panel's Findings**\n> a) If the panel finds that the declared system does **not** meet the definition of AGI, then OAI will immediately transfer to Microsoft all AGI Relevant Research used to develop the declared system unless OAI has not put a model or system resulting from the AGI Relevant Research to commercial use…\n> b) If the panel finds that the declared system **does** meet the definition of AGI:\n>   i) OAI need not transfer the AGI Relevant Research used to develop the evaluated model, and **AGI Research IP** (defined below) will no longer be subject to transfer requirements, though still may be transferred at OAI's discretion.\n>     (1) **\"AGI Research IP\"** means: (a) Confidential, non-public methods and know-how used in the research and development of a system that has been declared and verified as AGI; (b) Any related confidential, non-public artifacts, including confidential, nonpublic model training code and model weights (e.g., \"rail-free\" models built for safety testing purposes and models intended for internal deployment or research only); and (c) Any improvements to the above.\n>     (2) For clarity, \"AGI Research IP\" does **not** include: (a) **Model IP**: BOM for any publicly released model, including any model that meets the definition of AGI, including model architecture, model weights, inference code, and finetuning code. (b) **Supercomputer IP**: data center designs (including computer architecture), hardware (including silicon and networking) and software (including operating systems, compilers, runtimes, frameworks or middleware…). (c) **Previously transferred IP**: any Licensed IP that OAI has licensed to Microsoft, i.e., no retroactive application of AGI Research IP once transferred to Microsoft.\n>   ii) Mutual revenue sharing ceases.\n>   iii) IP SLA payments continue for failure to transfer Licensed IP.\n>   iv) True up is calculated as described above and the 10 equal annual installments period begins on the next 1/1/XX date. For example, if AGI is confirmed to be achieved and declared on 11/28/2029, then the first annual installment is due on 1/1/2030.\n>   v) None of the JDCA restrictions on OAI will apply to AGI Research IP or a system determined to be AGI or any successor systems or improvements to such system. … **Microsoft will have the right of first refusal to be OAI's commercialization partner for AGI** if OAI chooses to have a commercialization partner.\n>   vi) The parties will cooperate and coordinate on public disclosure, if any, of the panel's findings.\n>\n> **4) Argos 2 Restrictions Removed**\n> a) Subject to Section 4(b) below, the prohibitions on Microsoft pursuing AGI independent of OAI and on Microsoft's use of Licensed IP to pursue AGI are removed.\n> b) In any given year through the term, and only before a finding of AGI by the expert panel, Microsoft will not use Licensed IP to train a frontier model on a single, contiguous site with greater than the Approved Training Capacity. **\"Approved Training Capacity\"** means: i) [redacted]: up to [redacted]; ii) [redacted]: up to [redacted]; iii) [redacted]: up to [redacted]; iv) [redacted]: up to [redacted].\n> c) All other terms and conditions and rights and responsibilities associated with the terms \"AGI,\" \"Sufficient AGI,\" and \"Restricted Access Model Training Code\" are removed.\n>   i) For clarity, the definition of \"AGI\" will remain the same.\n>\n> **LICENSED IP**\n>\n> **1) Licensed IP**\n> a) The current definition of \"Licensed IP\" in the JDCA remains unchanged except for the inclusion of AGI and the exclusion of the following: ii) **Design IP**, defined as \"look and feel\" … (very high bar for exclusion beyond \"look and feel\"); iii) **Consumer hardware**; iv) **Excluded 3P Collaboration Product**; v) **AGI Research IP**.\n> b) OAI will use commercially reasonable efforts to remove **Non-Transferable Materials** from Licensed IP. … \"Non-Transferable Materials\" means: (1) Materials to the extent including customer pricing; product marketing and promotions; customer lists; customer contracts; customer targets; customer or user data; specialized vendor contracts and pricing; third-party supplier, distribution, and other collaboration plans, negotiations, and contracts; and employee compensation, partner, or customer contracts. (2) Materials to the extent including a product roadmap (to be defined in DA). (3) Materials to the extent including a business strategy (to be defined in DA). (4) Customer data … (5) Personal data of a party's users, customers, and partners. (6) Communications properly subject to attorney-client privilege. (7) Training data that is subject to third-party contractual or copyright restrictions … (8) AGI Relevant Research, properly withheld under AGI Section 1(a) above. (9) Models that are fine-tuned using customer data and that meet the definition of \"Excluded Fine-Tuned Customer Training Models\" under the JDCA.\n> c) **\"Excluded 3P Collaboration Product\"** means a product jointly developed by OAI and a customer, using Licensed IP, solely for that customer's commercial use or for sale in a limited commercial market.\n>   i) An Excluded 3P Collaboration Product must be: (1) solely for commercial use by the customer (e.g., customer-specific model); (2) sold by OAI as standalone solution that is expressly designed for and marketed to a limited and specific industry or use case; or (3) resold by OAI under 3P brand (e.g., *Contoso* or *Contoso for ChatGPT*, but not *ChatGPT Contoso*) …\n>\n> **IP TRANSFER** *[Sections 1, 2, 3 and 4 substantially redacted on the public copy. The visible material:]*\n>\n> **5. Inadvertent Transfers.** If OAI inadvertently transfers materials to Microsoft that should not have been transferred, Microsoft will return or delete those materials if OAI notifies Microsoft of the inadvertent transfer, or Microsoft identifies it independently, within a reasonable amount of time. …\n>\n> **6. Disclosure Team**\n> a. OAI will make available to the Disclosure team all information or materials necessary to verify that all Licensed IP has been transferred and track performance against the agreed SLAs. … Highly Sensitive materials include documents or information protected by attorney-client privilege, data of users or customers, third party personal information, sensitive security information, AGI Relevant Research, and information or technology that OAI deems dangerous from an AI safety perspective. The Disclosure Team will not have access to training data. …\n> b. For security reasons, no individual member of the Disclosure Team will have access across all such systems …\n> c. The Disclosure Team will primarily serve as an audit function …\n> d. The weekly cap on the number of requests …\n> e. There will be a pilot period of 90 days from the effective date of the amended and restated JDCA … The initial size of the Disclosure Team will consist of **10 Microsoft FTEs**.\n> f. Disclosure Team members will complete confidentiality training and will not share details regarding undisclosed or unshared Licensed IP …\n>\n> **OE PROGRAM, SUPPORT TEAM, SAFETY, OSS, GOVERNMENT**\n>\n> **1) OE program:** The program will be eliminated once: a) The dedicated OAI support team is fully staffed and trained. b) The disclosure team is scaled up while the OE team is scaled down, using SLAs as the metric: [60-day phase 1 → remove half OEs; second 60-day phase → remove remaining OEs; if OAI misses SLA targets, +30 days]. iv) If, after 1 year from the effective date of the amended and restated JDCA, the OE program has not been fully scaled down, then MSFT must choose whether to continue with either the OE program at a maximum of 20 members or the Disclosure Team for the remainder of the term of the JDCA. …\n> c) Once the OE program is fully wound down, parties will establish an **Embedded Engineering (EE) program** … initial size of **8** [5 embedded engineers covering inference, API sim-ship, safety; 2 with system access only (\"Rakesh\" engineers)] … iii) the expansion of the OE program set to occur in Jan 2026 under the JDCA will be removed from the agreement.\n>\n> **2) Support Team** — OAI will staff a Support Team which will initially consist of **11 FTEs**. … dedicated to helping MSFT commercialize Licensed IP and facilitating IP transfer.\n>\n> **3) Safety**\n> a) **DSB** will continue to conduct safety reviews before the first deployment of a new model by either party. … Microsoft and OAI will coordinate on all red teaming efforts for models anticipated to require DSB approval. …\n> b) OAI may, in its sole discretion and in good faith, declare a model, model component or model development technology unsafe for commercialization … Any such prohibitions or limitations will apply equally across all use cases, time periods, and geographic regions to all parties and third parties, including Microsoft, OAI non-profit, PBC, and all their affiliates. …\n> c) Microsoft will meet or exceed the same safety and security requirements for model deployments of Licensed IP as OAI uses for its own deployments, and each party will have the right to audit the other party's deployments …\n> d) AGI Research IP will be subject to Section 3(b) above but not Sections 3(a) or 3(c).\n>\n> **4) IP SLAs.** The payments owed for missing SLAs will not apply until 75 days from the effective date of the amended and restated JDCA. There will be three SLAs applicable to IP transfer in the areas of uptime of the automated transfer mechanisms, transfer of BoMs for models included in the IP Catalog, and transfers of IP for weekly high priority requests. … 1. Payments for missed SLAs are performance incentives and will not relieve OAI of any obligation to share Licensed IP. 2. Payments for missed SLAs are not remedies for breach of OAI's obligations to share Licensed IP, and Microsoft reserves its rights to pursue all remedies available for OAI's breach, at law and in equity. 3. OAI acknowledges and agrees that monetary damages would be inadequate to compensate Microsoft for breach of the IP transfer provisions. 4. In reliance on OAI's acknowledgment, **Microsoft promises not to pursue monetary damages for OAI's breach, only equitable relief.**\n>\n> **5) OSS.** OAI can open source the following models:\n> • **Older Models:** Through 2030, OpenAI can open-source any model that has been surpassed (to be defined in the DA) by a newer model within the same family or a subsequent family.\n> • **Smaller Models:** Through 2030, OAI may open-source any model that has at least 50% less model weights than the most recent publicly released frontier model within the same family or a subsequent family.\n> • **Non-Model Licensed IP:** OpenAI may open-source non-model Licensed IP in the following categories: standards and protocols, sales enablement code, reference applications, and GPU kernels. …\n> • Definition of model family: **Shared Architecture and Lineage.** …\n> • OSS-GPT will be considered part of the GPT5 family and part of Smaller Models …\n>\n> **6) Government**\n> 1. OAI may offer, deploy, and sell the OAI API for use in the following environments: a. Any US Federal Government customer (including Department of Defense) in any cloud environment that is the functional equivalent of Azure Government Secret or Azure Government Top Secret (e.g., AWS Secret Cloud and AWS Top Secret Cloud). b. United States Department of Defense customers in any cloud environment that is the functional equivalent of Azure Government (e.g. AWS GovCloud (US)) …\n> 2. Revenue share for any of the above will be calculated based on the amounts paid to the seller of record by the end customer instead of net revenue.\n> 3. … Microsoft will not unreasonably withhold the waiver.\n>\n> **OTHER TERMS**\n>\n> **1) Revenue share payment plan**\n> a) New mutual rev share payments: [redacted] in H2CY25, [redacted] in CY26, [redacted] in CY27, [redacted] in CY28, [redacted] in CY29, [redacted] in CY30.\n> b) **True up in June 2031, paid in 10 equal annual installments from 2031 to 2040.** Calculated as follows: Difference between NPV of JDCA mutual rev share payable of [redacted] and rev share actually paid, using [redacted] WACC.\n> c) IP SLA payments and revenue sharing from API exclusivity carveouts under the Government section above are not subject to the payment plan.\n>\n> **2) Stateless APIs** can be deployed by both companies on premises, subject to mutually agreed security requirements and definition of on premises.\n>\n> **3) Recapitalization**\n> a) **Microsoft preferred equity ownership post-recapitalization of 32.5% (fully diluted, prior to Golden Gate, Sakura and M&A deals).**\n> b) Microsoft to receive preferred rights equivalent to other OAI major investors, which would supersede the major decision construct that exists today (including current M&A and financing approvals).\n> c) Microsoft's consent to the proposed recapitalization structure subject to additional review and approval of definitive agreements and contingent upon all applicable regulatory approvals.\n>\n> **4) Extension of certain JDCA terms relating to Licensed IP**\n> a) **\"Extension IP\"** means Licensed IP as defined above provided that Extension IP does not include AGI Research IP or AGI Relevant Research regardless of whether AGI has been achieved as described above.\n> b) The licenses, disclosure and transfer obligations (e.g., Disclosure Team, SLAs), and restrictions on MSFT's commercial use (e.g., DSB approval, limitations on transfer, confidentiality) with respect to Extension IP, as modified by this Term Sheet, continue until **December 31, 2032 (\"IP Extension Period\")**, except: i) The approval rights in \"AGI\" Section 1(c) do not apply during the IP Extension Period. ii) OAI will transfer any Extension IP resulting from AGI Relevant Research that would have otherwise been required to be transferred under the terms of this Term Sheet using the processes described in this Term Sheet as if AGI had been attained.\n> c) The IP-related terms of the JDCA applicable to Licensed IP will apply to Extension IP except that any licenses to Extension IP that would have been exclusive will be **non-exclusive**.\n> d) No other terms of the JDCA apply to Extension IP. … neither party is obligated to share revenue that would otherwise be owed for use of Extension IP following December 31, 2030, even if AGI has not been achieved by that date.\n>\n> **5) Definitive Agreements.** The parties will use good faith efforts to executive [sic] definitive agreements consistent with this term sheet within 45 days following execution of the term sheet.\n>\n> **Signature Pages Follow**\n>\n> **OpenAI OpCo, LLC** — Signature: *(DocuSigned: Sam Altman, 63F8FF09BEF94A9)* — Name: Sam Altman — Title: Chief Executive Officer\n> **OpenAI, Inc.** — Signature: *(DocuSigned: Sam Altman, 63F8FF09BEF94A9)* — Name: Sam Altman — Title: Chief Executive Officer\n> **Microsoft Corporation** — Signature: *Amy Hood (Sep 11, 2025 12:26:27 PDT)* — Name: Amy Hood — Title: Chief Financial Officer\n>\n> **Exhibit A — Model BoM agreed 8.21**: Asset / Description / Transfer mechanism for Model snapshots (final lock model inference + training format; safety model; image and audio encoder/decoder), Inference code, Unreleased product code, Configs (engine manager + key system prompt), Model Reference document, Evals (3 capability or safety evaluation results), Golden example (input/output of a user interaction).\n>\n> **Exhibit B — SLA Table** (3 IP transfer areas):\n> 1. **Automation (code and models)** — SLA triggered if automatic transfer mechanism goes down/offline. SLA length 1 business day. Accrues after SLA ends. **$70K/Day/Infraction. Cap $225M Annual.**\n> 2. **BoM Delivery at Launch** for all models and products in model IP catalog — Trigger: Launch at public GA. SLA 2 business days; cure 2 business days. **$30K/Day/Infraction.**\n> 3. **Priority Requests** — Specific requested assets, capped at 70 points per week. SLA 5 business days; cure 2 business days. **$15K/Day.**\n\n## Commentary\n\nPX 212 is the actual paper of the **Watershed MOU** that anchored Robert Wu's Day 7 30(b)(6) deposition and is the structural fact behind plaintiffs' \"the AGI carveout is dissolving\" theme (see [[Key Themes]] §\"The Watershed MOU\"). Three load-bearing facts for plaintiffs land on the page itself: (1) the **expert panel decides AGI**, with Microsoft having a **right of first refusal to be OAI's commercialization partner for AGI** if a panel says yes — exactly the opposite of the 2018 Charter promise that AGI is excluded from any Microsoft license, which Brockman's 2020 \"fiduciary duty to humanity\" email had relied on; (2) Argos 2's **prohibitions on Microsoft pursuing AGI independent of OAI are removed**, capped only by an Approved Training Capacity that runs only \"before a finding of AGI by the expert panel\"; (3) **Microsoft's preferred equity is set at 32.5% fully diluted** post-recapitalization, with full \"major investor\" preferred rights superseding the existing JDCA major-decision construct — the structural form of the \"captured by Microsoft\" tweet [[Key Themes#\"Captured by Microsoft\" (statute-of-limitations fight)|theme]] now memorialized as a signed term sheet. The IP Extension Period to **Dec. 31, 2032** locks Microsoft's licenses and the Disclosure Team / SLAs / DSB regime into place beyond the original JDCA term, with the previously exclusive licenses converting to non-exclusive — the bargain Wu twice answered \"**I don't know**\" about quantitatively (`5/5/2026 Testimony @ ~15:13 PT`). For the [[Brockman Journal]] / \"fiduciary duty to humanity\" thread, this exhibit is the contemporaneous instrument that makes the 2026-vintage breach concrete.\n\n---\n*See also:* [[Day 7|Day 7 digest]] · [[Robert Wu]] · [[Key Themes]] · [[Brockman Journal]] · [[Greg Brockman]] · [[Quotes]]\n"}
{"exhibit_id": "PX-22", "exhibit": "PX 22", "party": "Plaintiffs", "type": "Compendium / multi-doc", "admitted_trial_day": "unknown", "pages": 22, "size_bytes": 2181798, "source_pdf": "PX-22.pdf", "bates": "22.001-22.022 (also 2024MUSK-0003112 to -0003133)", "pdf_url": "https://media.mts-in.com/PX-22.pdf", "body_markdown": "# PX 22 — OpenAI Inc.'s California Charitable Trust registration package (August 24, 2017)\n\n> **OpenAI Inc.'s formal registration as a California charitable trust** with the California Attorney General's Registry of Charitable Trusts — filed by **Adler & Colvin** on **August 24, 2017** (received by the AG's office **August 28, 2017**). The bundle contains the **CT-1 Form**, the Delaware Certificate of Incorporation, the OpenAI Bylaws, and the IRS 501(c)(3) determination letter — i.e., the consolidated paper trail of OpenAI's formal representations to charity regulators.\n\n## Document type\n**Compendium / multi-doc** — California Attorney General Registry of Charitable Trusts initial-registration package. 22 sub-Bates pages (`22.001`–`22.022`), also stamped `2024MUSK-0003112` through `2024MUSK-0003133`. Five components, in order:\n\n1. **California CT-1 — Initial Registration Form** (4 pages, 22.001–22.003): Charitable-trust registration filed under Cal. Gov. Code § 12585. Identifies OpenAI Inc., FEIN 81-0861541, address 335 Pioneer Way, Mountain View, CA 94041, phone (650) 387-6701. Officers listed: **Christopher Clark** (President/Director), **Jonathan Levy** (Sec/Treasurer), **Elon Musk** (Director), **Sam Altman** (Director). Date of formation: September 1, 2016. IRS exemption letter: November 3, 2016. Signed by Chris Clark, President.\n2. **Delaware Certificate of Incorporation of a Non-Stock Corporation** (2 pages, 22.004–22.005): Filed in Delaware on **December 8, 2015**. Article THIRD: \"**This Corporation shall be a nonprofit corporation organized exclusively for charitable and/or educational purposes within the meaning of section 501(c)(3) of the Internal Revenue Code of 1986. … The specific purpose of this corporation is to provide funding for research, technological development and distribution of technology related to artificial intelligence. The resulting technology will benefit the public and the corporation will seek to open source technology for the public benefit when applicable.**\" Article FIFTH: irrevocable dedication to 501(c)(3) purposes. Article SIXTH: no capital stock. Article SEVENTH: no members. Signed by **Jonathan Levy** as Authorized Officer.\n3. **Bylaws of OpenAI, Inc.** (~14 pages, 22.006–22.019): Table of Contents (Articles I–X): Registered Office and Principal Place of Business, Membership, Membership Rights, Board of Directors, Committees, Officers, Interested Director or Officer Transactions, Indemnification and Insurance, **Grants Administration** (Purpose of Grants, Board of Directors Oversight, Refusal/Withdrawal, Accounting, Restrictions on Contributions), Miscellaneous (Fiscal Year, Contracts, Amendments, Governing Law).\n4. **IRS Determination Letter** (1 page, 22.020): Dated **November 3, 2016**. Letter 947. Confirms OpenAI Inc.'s exemption under IRC § 501(c)(3), DLN 17052500330046, Effective Date of Exemption: **December 8, 2015**. Public Charity Status: 170(b)(1)(A)(vi). Contact person: **Eric Kaye** ID# 31612. Sent to OpenAI Inc. **c/o Adler & Colvin / Ingrid Mittermaier**, 235 Montgomery St, Ste 1220, San Francisco.\n5. **Adler & Colvin transmittal letter** (1 page, 22.022): On Adler & Colvin letterhead, dated **August 24, 2017**, signed by **Jorge Lopez**, sent via Federal Express to the California AG Registry of Charitable Trusts. Encloses: (1) originally-executed Form CT-1; (2) copy of Delaware Certificate of Incorporation; (3) copy of Bylaws; (4) copy of IRS Determination Letter; (5) **$25 check** for the initial registration fee.\n\n## Logistics\n- **Trial admission:** unknown.\n- **File size:** ~2.1 MB, 22 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `022.pdf`.\n\n## Transcribed text (selected highlights)\n\n> **CT-1 Form, on the activity description (22.002):**\n>\n> > \"**OpenAI, Inc. (\"OpenAI\") is a nonprofit artificial intelligence (\"AI\") scientific research organization. Its goal is to engage in research activities that advance digital intelligence in the way that is most likely to benefit humanity as a whole, unconstrained by a need to generate financial return. AI technology will help shape the 21st century, and OpenAI wants to help the world build safe AI technology and ensure that AI's benefits are as widely and evenly distributed as possible. To that end, OpenAI hopes to build AI as part of a larger community, and wants to openly share its plans and capabilities along the way.**\"\n\n> **Delaware Certificate, Article THIRD (22.004):**\n>\n> > \"This Corporation shall be a nonprofit corporation organized exclusively for charitable and/or educational purposes within the meaning of section 501(c)(3). … The specific purpose of this corporation is to provide funding for research, technological development and distribution of technology related to artificial intelligence. **The resulting technology will benefit the public and the corporation will seek to open source technology for the public benefit when applicable.**\"\n\n> **Bylaws, Article IX — Grants Administration (22.018):**\n>\n> > Section 3. **Refusal; Withdrawal.** The Board, in its absolute discretion, shall have the right to refuse to make any grants or contributions, or to render other financial assistance, for any or all of the purposes for which the funds are requested. …\n> > Section 5. **Restrictions on Contributions.** Unless otherwise determined by resolution of the Board in particular cases, this corporation shall retain complete control and discretion over the use of all contributions it receives, and all contributions received by this corporation from solicitations for specific grants shall be regarded as for the use of this corporation and not for any particular organization or individual mentioned in the solicitation. **This corporation may accept contributions earmarked by the donor exclusively for allocation to one or more foreign organizations or individuals only if the Board of this corporation: (a) has determined that the specific charitable activity for which the donation was made furthers this corporation's exempt purposes; (b) has approved in advance disbursements of funds to support such charitable activity; (c) retains discretion and control as to the use of the contributions received by this corporation; and (d) exercises appropriate supervision to ensure funds are actually spent for the intended purposes.**\"\n\n> **Adler & Colvin transmittal (22.022):**\n>\n> > August 24, 2017 — Attorney General, Registry of Charitable Trusts. **Re: Initial Registration for OpenAI, Inc. — California Corporation No. C3858313.**\n> >\n> > \"We enclose the initial registration of OpenAI, Inc., a Delaware nonprofit nonstock corporation which qualified to do business in California on January 7, 2016, pursuant to Government Code Section 12585. The following items are submitted: 1. Originally-executed Form CT-1; 2. Copy of Delaware Certificate of Incorporation; 3. Copy of Bylaws; 4. Copy of IRS Determination Letter; and 5. Our check for $25.00, payable to the Department of Justice, for the initial registration fee.\"\n> >\n> > Very truly yours, **Adler & Colvin** — by **Jorge Lopez**.\n\n## Commentary\n\nPX 22 is **the formal \"we are a public charity, and we mean it\" registration on the California Attorney General's books**. Three things land:\n\n1. **The mission language is now sworn to a charity regulator.** The same phrase plaintiffs treat as the bedrock representation — \"**unconstrained by a need to generate financial return**\" — is on the CT-1, attested to under penalty of perjury, in August 2017. That is **before** the Aug 21 / Aug 28 / Sept 12 / Sept 20 / Nov 6 entries in the [[Brockman Journal]]; before [[PX 158|Altman's \"I remain enthusiastic about the non-profit structure!\"]]; and before any decision had been made to convert to OpenAI LP. The temporal proximity matters: while Brockman is journaling about a $1B walk-away figure and \"morally bankrupt\" pivots in private, the same OpenAI is filing a public-charity registration with the California AG.\n2. **The Bylaws Article IX (Grants Administration)** is unusually specific for an early-stage nonprofit: variance powers, donor-restriction handling, and the foreign-grant supervision standard. Plaintiffs' charitable-trust theory rests in part on the proposition that OpenAI Inc.'s bylaws *themselves* contemplate strong donor-purpose protections — which the 2019 LP conversion then arguably circumvents.\n3. **Adler & Colvin / Ingrid Mittermaier** are the law firm of record on both the IRS determination letter and the California registration. Their involvement establishes that OpenAI Inc. had **specialist nonprofit counsel** at every formation step — context for Schizer's Day 9 testimony on what would have been \"**custom and practice**\" for retaining advisers (see [[Day 9|Day 9 digest]] §2d).\n\nThis exhibit pairs naturally with [[PX 16]] (the Delaware Certificate of Incorporation, which is the second component of this package, here viewed in the context of California-AG submission), [[PX 24]] (the 2018 OpenAI Charter), and the Form 990 series ([[PX 51]] / [[PX 52]] / [[PX 53]]).\n\n---\n*See also:* [[PX 16]] · [[PX 24]] · [[PX 51]] · [[PX 52]] · [[PX 53]] · [[Brockman Journal]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-222", "exhibit": "PX 222", "party": "Plaintiffs", "type": "Compendium / multi-doc", "admitted_trial_day": "unknown", "pages": 53, "size_bytes": 7146661, "source_pdf": "PX-222.pdf", "bates": "222.001-222.053 (also MSFT_MUSK000036519 - 000036571)", "pdf_url": "https://media.mts-in.com/PX-222.pdf", "body_markdown": "# PX 222 — OpenAI Inc.'s IRS Form 1023 application for 501(c)(3) tax-exempt status (September 1, 2016)\n\n> **OpenAI Inc.'s federal application for tax-exempt status as a 501(c)(3) public charity** — filed by **Adler & Colvin** (Ingrid Mittermaier signing) to the IRS in Covington, KY on **September 1, 2016**, with an **$850 application fee**. The federal companion to [[PX 22]] (the California Charitable Trust registration, filed almost exactly a year later). The Form 1023 packet includes the proposed budget, the COI policy, the Bylaws — and the **first sworn officer-compensation table**, including Sutskever at **$1,945,967**.\n\n## Document type\n**Compendium / multi-doc** — IRS Form 1023 application packet. **53 sub-Bates pages** (`222.001`–`222.053`), also stamped **`MSFT_MUSK000036519`** through **`MSFT_MUSK000036571`** (sourced from Microsoft's document production). Marked **HIGHLY CONFIDENTIAL**. Components:\n\n1. **Adler & Colvin transmittal letter** (2 pages, 222.001–002): Dated September 1, 2016; signed by **Ingrid Mittermaier**; addressed to IRS Internal Revenue Service, Attn: Extracting – Stop 312, Covington KY 41011-1454. Lists **11 enclosures**: (1) $850 check; (2) Form 1023 Checklist; (3) Form 1023; (4) Form 2848 Power of Attorney; (5) Form 5768 Election/Revocation of Election by an Eligible Section 501(c)(3) Organization to Make Expenditures to Influence Legislation; (6) Form 1023 Application for Recognition of Exemption; (7) Certificate of Incorporation, certified by the Delaware Division of Corporations; (8) Bylaws; (9) Proposed Budget; (10) Supplemental Response to Form 1023; (11) Conflict of Interest Policy; (12) Line of Credit Agreement. cc: OpenAI, Inc. (via email).\n2. **$850 check** to U.S. Department of the Treasury (1 page, 222.003): Drawn on Adler & Colvin City National Bank account, dated 8/11/2016, \"Client code: OPENAI.\"\n3. **IRS Form 1023** (the application proper, ~25 pages, 222.004–222.025+): Part II Organizational Structure (\"**Are you a corporation? Yes**\"); Part III Required Provisions in Your Organizing Document — locating the 501(c)(3) purposes-clause language at \"**Page 1, Paragraph Third**\" of the Delaware Cert and the dissolution clause at \"**Page 1, Paragraph Fifth**\"; Part IV Narrative Description of Your Activities; Part V Compensation and Other Financial Arrangements With Your Officers, Directors, Trustees, Employees, and Independent Contractors.\n4. **Bylaws of OpenAI, Inc.** (~14 pages, 222.024–037): Same Bylaws contained in [[PX 22]] — Articles I–X.\n5. **Supplemental Responses to Form 1023** (~5 pages, 222.038–042): Including the **first sworn compensation schedule** (see transcribed text below), and disclosed family/business relationships (Altman director, Levy Sec/Treasurer — both then employed by Y Combinator; Sutskever has a business interest in YC).\n6. **Conflict of Interest Policy + disclosure form** (~8 pages, 222.044–051): Standard nonprofit COI policy plus blank annual-disclosure form.\n7. **FedEx shipping label** (1 page, 222.053): Priority Overnight, FedEx Ship Manager, Sept 2, 2016 10:30 AM ET, sent from Adler & Colvin to the IRS Covington address.\n\n## Logistics\n- **Trial admission:** unknown.\n- **File size:** ~6.8 MB, 53 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `222.pdf`.\n- **Production source:** Microsoft (`MSFT_MUSK000036519`–`000036571`). Notable: this OpenAI tax-exemption document came in via Microsoft's discovery production, not OpenAI's own — an artifact of the documents Microsoft had reviewed during its 2019 due diligence.\n\n## Transcribed text\n\n> **Adler & Colvin cover letter (222.001):**\n>\n> > September 1, 2016 — Via Federal Express. Internal Revenue Service. **Re: OpenAI, Inc. — Employer Identification Number 81-0861541.**\n> >\n> > \"We enclose the tax exemption application of OpenAI, Inc. ('OpenAI'), a Delaware nonprofit nonstock corporation, for your review. **OpenAI seeks a determination that it is a charitable organization described in Section 501(c)(3) of the Internal Revenue Code, and a publicly supported organization described in Section 509(a)(1) and 170(b)(1)(A)(vi).**\"\n>\n> Signed: **Ingrid Mittermaier**, Adler & Colvin.\n\n> **Form 1023 — Part III Required Provisions (222.010):** OpenAI's organizing document is the Delaware Certificate of Incorporation. Section 501(c)(3) purposes clause: located at **Page 1, Paragraph Third**. Dissolution clause confirming distribution upon dissolution to another 501(c)(3): **Page 1, Paragraph Fifth**.\n\n> **Supplemental Response — Compensation table (222.040):**\n>\n> > **Question 1b: Compensation and Other Information Regarding Employees.**\n> >\n> > | Name of Employee | Title | Compensation |\n> > |---|---|---|\n> > | **Ilya Sutskever** | Research Director | **$1,945,967** |\n> > | **Ian Goodfellow** | Technical Staff | $875,000 |\n> > | **Pieter Abbeel** | Technical Staff | $550,000 |\n> > | **Man Wai Vicki Cheung** | Technical Staff | $307,526 |\n> > | **John Schulman** | Technical Staff | $284,610 |\n> >\n> > \"Note the compensation listed above for each employee includes a fixed bonus payment. **Fixed bonuses were negotiated at arm's length and paid to recruit these top scientific researchers from other, more lucrative positions at prominent technology companies.**\"\n> >\n> > \"The mailing address for all employees is: **335 Pioneer Way, Mountain View, California 94041**.\"\n\n> **Supplemental Response — Question 2a/c: Family/Business Relationships with/among Key Personnel (222.040):**\n>\n> > \"**Mr. Altman, director, and Mr. Levy, Secretary and Treasurer, are both employed by Y Combinator. Mr. Sutskever, Research Director, has a business interest in Y Combinator.**\"\n\n> **Supplemental Response — Question 3a: Qualifications and Duties of Directors and Officers (222.040):**\n>\n> > **i. Chris Clark, President, Director, and Chief Operating Officer.**\n> >\n> > Duties and Hours: \"As President, Mr. Clark presides at all meetings of the Board of Directors and, subject to control of the Board, generally supervises, directs, and controls the business and other officers of the corporation. … As Chief Operating Officer ('COO'), **Mr. Clark oversees day-to-day operations of OpenAI and is responsible for all non-technical activities. Mr. Clark is compensated for his services as COO and spends 20 hours a week to accomplish his duties as COO.**\"\n> >\n> > Qualifications: \"Chris Clark was Vice President of North America at Canonical, maker of the world's most popular open source operating system (Ubuntu). More recently, **he was Mayor of Mountain View** and still serves on the city council. Mr. Clark has a BA in Political Science from Stanford University.\"\n\n## Commentary\n\nPX 222 is **the federal-tax-exemption parallel to [[PX 22]]** — together they form the consolidated paper trail of OpenAI Inc.'s formal, sworn representations to the federal IRS (Sept 2016) and the California AG (Aug 2017) that it was a 501(c)(3) public charity. Three load-bearing observations:\n\n1. **The compensation table on 222.040 is the first sworn record of officer pay** at OpenAI — predating the [[PX 51|2016 Form 990]] by months. **Sutskever at $1,945,967** is the same figure plaintiffs invoke in the 2016 990 ($900K base + $1M bonus, see [[PX 51]] Schedule J). The fact that the Form 1023 supplemental discloses these numbers — including the disclosure that the bonuses were \"negotiated at arm's length\" and \"paid to recruit … from other, more lucrative positions at prominent technology companies\" — is itself part of plaintiffs' framing that OpenAI was always already operating like a for-profit on the compensation side, even while presenting publicly as a nonprofit.\n2. **Altman/Levy/Sutskever's Y Combinator entanglement is disclosed up front.** The supplemental response in 222.040 acknowledges that two directors and the senior research lead were affiliated with YC — a fact that becomes load-bearing later when [[PX 71]] (the 2016 fiscal-sponsorship structure) and [[PX 60]] (the 2014–2017 DAF flow into \"YC ORG\") show how grants flowed through that relationship.\n3. **The document came from Microsoft's production.** The MSFT_MUSK Bates run means this packet was reviewed by Microsoft during due diligence — Microsoft had OpenAI Inc.'s 501(c)(3) application in hand before/during the 2019 LP investment. Plaintiffs use that fact to rebut any defense argument that Microsoft was unaware of the nonprofit's representations or restrictions.\n\nPair this exhibit with [[PX 22]] (CA AG registration, August 2017) and [[PX 16]] (Delaware Cert of Incorporation, December 2015) for the full formation paper trail.\n\n---\n*See also:* [[PX 16]] · [[PX 22]] · [[PX 24]] · [[PX 51]] · [[PX 52]] · [[PX 53]] · [[PX 60]] · [[PX 71]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-229", "exhibit": "PX 229", "party": "Plaintiffs", "type": "Demonstrative", "admitted_trial_day": "unknown", "pages": 21, "size_bytes": 1722596, "source_pdf": "PX-229.pdf", "bates": "229.010-229.030+", "pdf_url": "https://media.mts-in.com/PX-229.pdf", "body_markdown": "# PX 229 — Microsoft \"OpenAI on Azure Big Compute\" internal pitch deck (2016)\n\n> A **MICROSOFT CONFIDENTIAL** internal slide deck pitching the OpenAI partnership to Microsoft leadership in the run-up to the September 2016 contract renewal — the document that captures Satya Nadella's **first** approach to Sam Altman about migrating OpenAI off AWS and onto Azure.\n\n## Document type\n**Demonstrative / internal slide deck** (\"MICROSOFT CONFIDENTIAL\"). PowerPoint export to PDF, ~21 pages. Bates 229.010 onward.\n\n## Logistics\n- **Trial admission:** unknown.\n- **File size:** ~1.7 MB, 21 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `229.pdf`.\n\n## Transcribed text\n\n> **Slide: OpenAI on Azure Big Compute — Overview, Recommendations and Models** (MICROSOFT CONFIDENTIAL).\n>\n> **Slide: OpenAI Overview**\n> - OpenAI is a non-profit artificial intelligence research company. Their goal is to advance digital intelligence in the way that is most likely to benefit humanity as a whole, **unconstrained by a need to generate financial return**.\n> - **Elon Musk** (founder of Tesla) and **Sam Altman** (President of YCombinator) are founders and co-chairs of OpenAI. Additionally backed by Peter Thiel and Reid Hoffman.\n> - Currently create their models, algorithms and cutting edge research on **AWS** as part of a deal that provides them **$50M in GPU compute for $10M in committed funds**. **Deal is coming for renewal on 9th September 2016 at $10M / $60M in equivalent compute.**\n> - First research organization in the world with access to **NVIDIA's DGX-1** (Supercomputer in a box) offering with Pascal GPUs with additional **10k Maxwell GPUs donated**.\n> - **Satya approached Sam** with the desire to work closer together given our focus on AI. AWS garners a lot of mindshare currently from OpenAI.\n>\n> **Slide: Benefits to Microsoft**\n> - **Thought Leadership:** Unique opportunity for MS Corp to attach and support high profile, non profit organization driving innovation in the AI segment — Not many deals or opportunities like this.\n> - **Momentum:** Announce the partnership at Ignite & get momentum before AWS's competitive GPU offering — their release is imminent.\n> - **Driving Azure Usage:**\n>   - \"Halo effect\" for upcoming GPUs launch via partnership with cutting edge technology company and thought leaders in the Deep Learning Space.\n>   - Enable Azure to recruit net-new audience of next generation of developers & start-ups.\n>   - Accelerate our development of DL in Azure by having OpenAI develop their latest tools, algorithms & libraries on Azure and accelerate the virtuous cycle of CNTK development.\n>   - Drive increased bursting usage from the DL community by attaching to the DGX-1 Device that NVIDIA provides.\n>   - Grow algorithmic usage of Tensor Flow and Caffe in our solutions — We'd have immediate customers.\n> - **Attached Azure Revenue:** Generate pull through revenue for additional services in Azure, hereby creating a \"Better together\" story for deep learning solutions in Azure.\n>   - Standard compute for infrastructure based services such as web servers as part of the Deep Learning architecture.\n>   - Strong Interest in Azure Batch for their deployment & scheduling DNN jobs across GPU based nodes.\n>   - Azure Storage for storage of their input data, models.\n>\n> **Slide: Deal Ask**\n> - Their ask is to match their next proposal to AWS, if we do they'll switch:\n>   - **3 year deal worth the equivalent of $60M in AWS GPU** [remainder of the financial terms continue across subsequent slides — not fully transcribed].\n>\n> [Remaining slides cover Azure model recommendations, technical architecture, and pricing scenarios. Not all extractable from the layout text.]\n\n## Commentary\n\nPX 229 is **the smoking-gun document on Microsoft's earliest interest in OpenAI** — internal, confidential, and predating the [[PX 877|July 2019 OpenAI-Microsoft announcement]] by **nearly three years**. Three things land:\n\n1. **\"Satya approached Sam\"** in 2016 — Microsoft, not OpenAI, initiated the relationship that would eventually produce the LP investment, the JDCA, and Microsoft's ~$200B economic stake. This sequence undercuts any defense narrative that Microsoft was a passive recipient of OpenAI's overtures.\n2. **The 2016 AWS context** — OpenAI was on AWS at $50M-for-$10M-cash terms, with a renewal coming up at $60M-for-$10M-cash, and was already running on the **first NVIDIA DGX-1 ever delivered** (see [[PX 388]]) plus 10K donated Maxwell GPUs.\n3. **Microsoft framed the partnership in commercial terms** — \"Driving Azure Usage,\" \"Attached Azure Revenue,\" \"Halo effect,\" \"Better together\" — language plaintiffs use to argue that Microsoft's interest was always commercial-strategic, regardless of OpenAI's nonprofit packaging.\n\nThis deck connects to the timeline laid down by [[PX 70]] (April 2016 Musk–Huang DGX-1 thread), [[DX 556]] (Aug 2016 \"$50M compute donation\"), and [[PX 90]] (June 2017 \"Re: The 10,000\" Microsoft / Azure GPU ask).\n\n---\n*See also:* [[PX 70]] · [[PX 90]] · [[PX 388]] · [[PX 877]] · [[DX 556]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-233", "exhibit": "PX 233", "party": "Plaintiffs", "type": "Email", "admitted_trial_day": "Day 3 (April 29, 2026)", "uploaded_box_pt": "2026-04-29T15:09:20", "uploader": "Someone", "pages": 2, "size_bytes": 310078, "source_pdf": "PX-233.pdf", "pdf_url": "https://media.mts-in.com/PX-233.pdf", "body_markdown": "# PX 233 — April 23, 2018 Zilis → Musk \"AI updates\" briefing\n\n> Shivon Zilis's two-page status briefing to Elon Musk covering the post-board-resignation OpenAI fundraising plan, the Cerebras chip timeline, the Dota 5v5 progress, and Zilis's own time-allocation across OpenAI/Neuralink/Tesla — written about two months after Musk's February 20, 2018 resignation from the OpenAI nonprofit board.\n\n## Document type\n**Email thread, plain text, two messages.** Musk's two-word reply (\"Ok by me\") on top of Zilis's April 23, 2018 1:49 AM PDT email; with a longer March 25, 2018 11:03 AM Zilis email further down. Recipients: Shivon Zilis <shivon@neuralink.com>, Sam Teller <steller@spacex.com>. Bates 2024MUSK-0005604–5605.\n\n## Logistics\n- **Trial admission:** Day 3 (April 29, 2026) — used during the Musk cross block to bridge the September 2017 control fight and the February 2018 board exit, and to anchor the \"Tesla AI play\" theme. No prior wiki reference; admission day inferred from Box upload date.\n- **Box upload:** 2026-04-29 15:09:20 PT — Day 3 mid-afternoon batch (clustered with PX 157, PX 296, PX 355).\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~303 KB, 2 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `233.pdf`.\n\n## Transcribed text\n\n> **From:** Elon Musk <erm@spacex.com>\n> **To:** Shivon Zilis <shivon@neuralink.com>\n> **Cc:** Sam Teller <steller@spacex.com>\n> **Subject:** Re: AI updates\n> **Date:** Mon, 23 Apr 2018 02:06:23 -0700\n>\n> Ok by me\n\n> **From:** Shivon Zilis <shivon@neuralink.com>\n> **To:** Elon Musk <erm@spacex.com>\n> **Cc:** Sam Teller <steller@spacex.com>\n> **Subject:** AI updates\n> **Date:** Mon, Apr 23, 2018 at 1:49 AM\n>\n> Updated info per a conversation with Altman. You're tentatively set to speak with him on Tuesday.\n>\n> **Financing:**\n> - He confirmed again that they are definitely not doing an ICO but rather equity that has a fixed maximum return.\n> - Would be a rather unique subsidiary structure for the raise which he wants to walk you through.\n> - Wants to move within 4-6 week on first round (probably largely Reid money, potentially some corporates).\n>\n> **Tech:**\n> - Says Dota 5v5 looking better than anticipated.\n> - The sharp rise in Dota bot performance is apparently causing people internally to worry that the timeline to AGI is sooner than they'd thought before.\n> - Thinks they are on track to beat Montezuma's Revenge shortly.\n>\n> **Time allocation:**\n> - I've reallocated most of the hours I used to spend with OpenAI to Neuralink and Tesla. This naturally happened with you stepping off the board and related factors — but if you'd prefer I pull more hours back to OpenAI oversight please let me know.\n> - Sam and Greg asked if I'd be on their informal advisory board (just Gabe Newell so far), which seems fine and better than the formal board given potential conflicts? If that doesn't feel right let me know what you'd prefer.\n>\n> **From:** Shivon Zilis <shivon@neuralink.com>\n> **To:** Elon Musk <erm@spacex.com>\n> **Cc:** Sam Teller <steller@spacex.com>\n> **Subject:** AI updates\n> **Date:** Sun, Mar 25, 2018 at 11:03 AM\n>\n> **OpenAI**\n>\n> Fundraising:\n> - No longer doing the ICO / \"instrument to purchase compute in advance\" type structure. Altman is thinking through an instrument where the 4-5 large corporates who are interested can invest with a return capped at 50x if OpenAI does get to some semblance of money-making AGI. They apparently seem willing just for access reasons. He wants to discuss with you in more detail.\n>\n> Formal Board Resignation:\n> - You're still technically on the board so need to send a quick one liner to Sam Altman saying something like \"With this email I hereby resign as a director of OpenAI, effective Feb 20th 2018\".\n>\n> Future Board:\n> - Altman said he is cool with me joining then having to step off if I become conflicted, but is concerned that others would consider it a burned bridge if I had to step off. I think best bet is not to join for now and be an ambiguous advisor but let me know if you feel differently. They have Adam D'Angelo as the potential fifth to take your place, which seems great?\n>\n> **TeslaAI**\n>\n> Andrej has three candidates in pipeline, may have 1-2 come in to meet you on Tuesday. He will send a briefing note about them. Also, he's working on starter language for a potential release that will be ready to discuss Tuesday. It will follow the \"full-stack AI lab\" angle we talked about but, if that doesn't feel right, please course correct... is tricky messaging.\n>\n> **Cerebras**\n>\n> Chip should be available in August for them to test, and they plan to let others have remote access in September. The Cerebras guy also mentioned that a lot of their recent customer interest has been from companies upset about the Nvidia change in terms of service (the one that forces companies away from consumer grade GPUs to enterprise Pascals / Voltas). Scott Gray and Ilya continue to spend a bunch of time with them\n>\n> CONFIDENTIALITY NOTICE: The contents of this email message and any attachments are intended solely for the addressee(s) and may contain confidential and/or privileged information and may be legally protected from disclosure.\n\n## Commentary\n\nPX 233 fixes several key facts in early 2018: (i) it confirms Musk's **February 20, 2018 board resignation date** and shows Altman handling Zilis's transition with care (\"step off if I become conflicted\"); (ii) it documents that **two months after Musk left the OpenAI board**, Altman was already pitching him on a \"**fixed maximum return**\" investor structure — i.e., the capped-profit framework that became the August 2018 LP term sheet ([[DX 827]]) — and that the round was being lined up with Reid (Hoffman) and \"some corporates\"; (iii) it puts Musk's own time-allocation language into the record — \"I've reallocated most of the hours I used to spend with OpenAI to Neuralink and Tesla\" — alongside parallel **TeslaAI** hiring activity (Andrej Karpathy \"has three candidates in pipeline\"). Defense uses the latter cluster to advance the parallel-AGI-play theme (see [[Key Themes]] § \"Tesla / xAI as Musk's own AGI play\" and [[DX 766]] on the Feb 2018 Karpathy poach list). Plaintiffs use the same email to argue that Musk understood and was kept in the loop on the for-profit structure as it was being designed. The Cerebras paragraph is also useful to plaintiffs' Cerebras-conflict-of-interest theory built around Brockman's earlier Cerebras stock purchase (see [[Key Themes]]).\n\n---\n*See also:* [[Day 3|Day 3 digest]] · [[Shivon Zilis]] · [[Sam Altman]] · [[DX 827]] · [[DX 766]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-236", "exhibit": "PX 236", "party": "Plaintiffs", "type": "Email + attachment (term sheet)", "admitted_trial_day": "Day 3 (April 29, 2026)", "uploaded_box_pt": "2026-04-29T15:09:21", "uploader": "Someone", "pages": 6, "size_bytes": 342184, "source_pdf": "PX-236.pdf", "pdf_url": "https://media.mts-in.com/PX-236.pdf", "body_markdown": "# PX 236 — Aug 31, 2018 Altman → Musk \"OpenAI term sheet\" + LP Summary Term Sheet\n\n> Sam Altman's August 31, 2018 cover email to Musk attaching the **OpenAI LP Summary of Principal Terms** — the original \"purple-box\" warning (\"**view any investment in OpenAI LP in the spirit of a donation**\"), the $500M / 100x initial raise, the $100B employee pool cap, and the \"fiduciary duties of the Nonprofit Board of Directors flow exclusively to the Nonprofit, not to the Limited Partners.\"\n\n## Document type\n**Email + attachment (slide-deck-style term sheet).** Page 1: Altman → Musk cover email (cc Sam Teller, Shivon Zilis), 11:30 AM CEST. Pages 2–6: the OpenAI LP Summary Term Sheet PDF — table-formatted, with the famous purple-bordered \"IMPORTANT WARNING\" box. Subject \"OpenAI term sheet.\" Bates SPX-001642–001647.\n\n## Logistics\n- **Trial admission:** Day 3 (April 29, 2026). No prior wiki reference; admission day inferred from Box upload date. Used in tandem with [[DX 827]] / [[DX 828]] on Day 4 for the [[Key Themes#The 2018 \"purple box\" term sheet|purple-box]] cross.\n- **Box upload:** 2026-04-29 15:09:21 PT — Day 3 mid-afternoon plaintiffs' batch (clustered with [[PX 157]], [[PX 158]], PX 105).\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~334 KB, 6 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `0202.pdf`-numbered (filename: `0202.pdf` no — actual file `236.pdf`). Public access to admitted exhibit.\n\n## Transcribed text\n\n> **From:** Sam Altman <shga@ycombinator.com>\n> **To:** Elon Musk <erm@spacex.com>\n> **Cc:** Sam Teller <steller@spacex.com>, Shivon Zilis <shivon@openai.com>\n> **Subject:** OpenAI term sheet\n> **Date:** Fri, 31 Aug 2018 11:30:15 +0200\n> **Importance:** Normal\n> **Attachments:** OpenAI_LP_-_Summary_of_Principal_Terms.pdf\n>\n> Elon--\n>\n> Please see attached, look forward to feedback.\n>\n> Also, my current thought is that I won't take any equity. I'm not doing this for the money anyway, and I like the idea of being completely unconflicted and just focused on the best outcome for the world. If it appeared at some point we weren't going to build AGI but were going to build something valuable, then maybe I'd want equity then.\n>\n> Sam\n\n---\n\n> **OpenAI LP — Summary Term Sheet**\n>\n> **IMPORTANT WARNING [purple box]**\n>\n> **Investing in OpenAI LP (the Partnership) is a *high-risk investment*. Investors could lose their capital contribution and not see any return. It would be wise to view any investment in OpenAI LP in the spirit of a donation, with the understanding that it may be difficult to know what role money will play in a post-AGI world.**\n>\n> As described herein and in the Limited Partnership Agreement, the Partnership exists to advance OpenAI Inc's mission of ensuring that safe artificial general intelligence is developed and benefits all of humanity. The General Partner's duty to this mission and the principles advanced in the OpenAI Inc Charter take precedence over any obligation to generate a profit. The Partnership may never make a profit, and the General Partner is under no obligation to do so. The General Partner is free to re-invest any or all of the Partnership's cash flow into research and development activities and/or related expenses without any obligation to the Limited Partners. See Section 6.4 of the Limited Partnership Agreement for additional details.\n>\n> **Mission**\n> OpenAI's mission is to ensure that artificial general intelligence — by which we mean highly autonomous systems that outperform humans at most economically valuable work — (AGI), when developed, is safe and benefits all of humanity. We will attempt to directly build safe and beneficial AGI, but will also consider our mission fulfilled if our work aids others to achieve this outcome.\n>\n> **Corporate Structure**\n> OpenAI LP will be a for-profit Delaware Limited Partnership managed by its General Partner, a single-member Delaware LLC controlled by OpenAI, Inc. (the Nonprofit)'s Board of Directors. At all times, no more than a minority of the Nonprofit's Board of Directors will be holders of any economic interest in OpenAI LP or in the employee holdings entity (described below).\n>\n> **Funding / Revenue**\n> *Initial Capitalization.* OpenAI LP will initially be capitalized by a contribution of assets from the Nonprofit. The Nonprofit will get an interest equivalent to that of a Limited Partner in the initial raise that is consistent with the value of its capital contribution.\n>\n> *Limited Partner Interests.* We will sell preferred capped Limited Partner Interests redeemable (on a first in, first out basis) for a multiple of the purchase price (adjusted for inflation as described below). Redemption will commence either (a) if and when we successfully create AGI capable of generating the requisite returns and the General Partner, in its sole and absolute discretion, determines that redemptions will begin, or (b) at a mutually agreeable pre-AGI alternative.\n>\n> Specifically, the fundraising rounds will be as follows:\n> 1. Initial raise: $500M, 100x target redemption\n> 2. Employee pool: $100B capped target redemption\n>    - Actual redemption amount will be based on the number of employee LP Interests issued, which will be determined (subject only to the aggregate cap) by the General Partner in its sole and absolute discretion.\n>    - The General Partner may increase the target redemption of the employee pool in its sole and absolute discretion.\n> 3. Second raise: $10+B, 15x expected target redemption.\n> 4. There may be additional fundraising rounds and/or an additional employee pool, though the interests of employee and/or investor Limited Partners cannot be diluted absent approval of a majority-in-interest of the affected class of Limited Partner (employee / investor).\n>\n> *Distributions.* (i) First, distributions to initial-raise investors in proportion to their respective capital contributions until each investor has received an amount equal to its aggregate capital contributions, with such distributions counting against the investors' target redemption amount. (ii) Next, 25% to initial-raise investors and employees in proportion to target redemption amounts and 75% to second-raise investors in proportion to capital contributions until they have received an amount equal to their contributions; thereafter, all future payments in proportion to target redemption amounts; distributions continue until all investors and employees have received their respective target redemption amounts. (iii) **Any amounts in excess of the target redemption amounts shall be distributed to or for the benefit of the Nonprofit in its capacity as a Limited Partner.**\n>\n> Inflation rate (years 1–10): U.S. CPI or alternative U.S. governmental index. Year 11+: greater of inflation rate or simple LIBOR + 2%.\n>\n> *Revenue.* OpenAI LP may generate revenue (interest, dividends, capital gains, royalties). To the extent it commercializes technology subject to UBIT, OpenAI LP will create taxable \"corporate blocker\" subsidiaries.\n>\n> *Tax Distributions.* OpenAI LP will make customary tax distributions in respect of net taxable income.\n>\n> **Fiduciary Duties**\n> OpenAI LP exists to advance our mission using the principles expounded in OpenAI's Charter. … Our duty to these principles and the advancement of our mission takes precedence over any obligation to generate a profit. We may never make a profit, and we are under no obligation to do so. … **The fiduciary duties of the Nonprofit Board of Directors flow *exclusively* to the Nonprofit, not to the Limited Partners.**\n>\n> **Employee Comp.**\n> Employees granted profit interests in an employee holding vehicle (OpenAI Holdings, L.P.); 6-year vest, 0% year 1, 20%/year for next 5; 1-year transfer cliff; max 25% of vested per year; Section 83(b) elections.\n>\n> **Employee Holding Company**\n> OpenAI Holdings, L.P. issues mirror profits interests; employees receive W-2s from OpenAI LP and K-1s from OpenAI Holdings.\n>\n> **Limitations Re: Transfer of Interests**\n> Annual transfer windows; LP-to-LP only absent GP consent; securities-law and PTP-tax-law caps.\n>\n> **Information Rights**\n> Annual financial statements within 90 days of FY end; payment confidentiality; carve-outs for required governmental reporting (e.g., Form 990 disclosure).\n>\n> **Strategic Investment Agreements**\n> OpenAI LP may enter strategic investment agreements with companies that become Limited Partners, \"subject only to the caveat that those agreements must not clearly subvert the redemption waterfall described above nor clearly detract from the advancement of our mission.\"\n\n## Commentary\n\nPX 236 is the *plaintiffs'* version of the same exhibit defense calls [[DX 827]] — the August 31, 2018 OpenAI LP term sheet — and is the centerpiece of the [[Key Themes#The 2018 \"purple box\" term sheet|\"purple box\"]] dispute. Plaintiffs lean on the language that the GP's duty to mission \"takes precedence over any obligation to generate a profit,\" that \"any amounts in excess of the target redemption amounts shall be distributed to or for the benefit of the Nonprofit,\" and that nonprofit-board fiduciary duties run \"*exclusively* to the Nonprofit\" — to argue the LP was *legally* a charitable vehicle. Defense uses the same purple box (\"**spirit of a donation**\" + \"high-risk investment\") to argue Musk had constructive notice of a for-profit redemption-waterfall structure when it was sent to him directly. Altman's \"**I won't take any equity… I'm not doing this for the money anyway**\" cover line is a separate plaintiff hook — used to frame Altman's later 2.0 equity discussion as a reversal. Cross-reference: see [[DX 827]], [[DX 828]] (Birchall to Zilis: \"plain vanilla for-profit structure\"), and Birchall on cross — \"I did not look closely at the term sheet\" `(042926TT.txt:6019)`.\n\n---\n*See also:* [[Day 3|Day 3 digest]] · [[Sam Altman]] · [[DX 827]] · [[DX 828]] · [[PX 24]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-239", "exhibit": "PX 239", "party": "Plaintiffs", "type": "Email", "admitted_trial_day": "Day 3 (April 29, 2026)", "uploaded_box_pt": "2026-04-29T15:09:18", "uploader": "Someone", "pages": 4, "size_bytes": 294170, "source_pdf": "PX-239.pdf", "pdf_url": "https://media.mts-in.com/PX-239.pdf", "body_markdown": "# PX 239 — Mar 6, 2019 Altman → Musk, capped-profit launch draft (\"Anything to add/edit?\")\n\n> Sam Altman's email to Musk (cc Sam Teller, Shivon Zilis) attaching the draft public-launch blog post for **OpenAI LP** — the \"capped-profit\" structure, the \"purple box\" first investor disclosure, and the news that \"we are now discussing a multi-billion dollar investment.\"\n\n## Document type\n**Email, with multi-page draft blog post attached.** Sent Wed, 6 Mar 2019 15:12:20 -0800 by Altman to erm@spacex.com, cc steller@spacex.com and shivon@neuralink.com. Bates 2024MUSK-0005585–88.\n\n## Logistics\n- **Trial admission:** Day 3 (April 29, 2026).\n- **Box upload:** 2026-04-29 15:09:18 PT — Day-3 mid-afternoon batch (clustered with [[PX-157]], [[PX-159]]).\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~287 KB, 4 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `239.pdf`.\n\n## Transcribed text\n\n> **From:** Sam Altman <shga@ycombinator.com>\n> **To:** Elon Musk <erm@spacex.com>\n> **Cc:** Sam Teller <steller@spacex.com>, Shivon Zilis <shivon@neuralink.com>\n> **Subject:** OpenAI\n> **Date:** Wed, 6 Mar 2019 15:12:20 -0800\n> **Importance:** Normal\n>\n> Elon —\n>\n> Here is a draft post we are planning for Monday. Anything to add/edit?\n>\n> **TL;DR:**\n> * We've created the capped-profit company and raised the first round, led by Reid and Vinod.\n> * We did this is a way where all investors are clear that they should never expect a profit (see purple box below).\n> * We made Greg chairman and me CEO of the new entity.\n> * We have tested this structure with potential next-round investors and they seem to like it.\n>\n> Speaking of the last point, we are now discussing a multi-billion dollar investment which I would like to get your advice on when you have time. Happy to come see you some time you are in the bay area.\n>\n> Sam\n>\n> ---\n>\n> **We've created OpenAI LP, a new \"capped-profit\" company that allows us to rapidly increase our investments in compute and talent while including checks and balances to actualize our mission.**\n>\n> Our mission is to ensure that artificial general intelligence (AGI) benefits all of humanity, primarily by attempting to build safe AGI and share the benefits with the world.\n>\n> Due to the exponential growth of compute investments in the field, we've needed to scale much faster than we'd planned when starting OpenAI. We expect to need to raise many billions of dollars in upcoming years for large-scale cloud compute, attracting and retaining talented people, and building AI supercomputers.\n>\n> We haven't been able to raise that much money as a nonprofit, and though we considered becoming a for-profit, we were afraid that doing so would mean giving up our mission. Instead, we created a new company, OpenAI LP, as a hybrid for-profit and nonprofit — which we are calling a \"capped-profit\" company.\n>\n> The fundamental idea of OpenAI LP is that investors and employees can get a fixed return if we succeed at our mission, which allows us to raise investment capital and attract employees with startup-like equity. But any returns beyond that amount — and if we are successful, we expect to generate orders of magnitude more value than we'd owe to people who invest in or work at OpenAI LP — are owned by the original OpenAI Nonprofit entity.\n>\n> Going forward (in this post and elsewhere), \"OpenAI\" refers to OpenAI LP (which now employs most of our staff), and the original entity is referred to as \"OpenAI Nonprofit\".\n>\n> *[OpenAI team and their families at our November 2018 offsite — image]*\n>\n> **The mission comes first**\n>\n> We've designed OpenAI LP to put our overall mission — ensuring the creation and adoption of safe and beneficial AGI — over generating returns for investors.\n>\n> To minimize conflicts of interest with the mission, OpenAI LP's primary fiduciary obligation is to advance the aims of the OpenAI Charter, and the company is controlled by OpenAI Nonprofit's board. All investors and employees sign agreements that OpenAI LP's obligation to the Charter always comes first, even at the expense of some or all of their financial stake.\n>\n> *[Big purple box — image]* \"Our employee and investor paperwork starts like this. The general partner refers to OpenAI Nonprofit (whose official name is 'OpenAI Inc'); limited partners refers to investors and employees.\"\n>\n> Only a minority of board members can hold financial stakes in the partnership. Furthermore, only board members without such stakes are allowed to vote on decisions where the interests of limited partners and the nonprofit's mission may conflict — including any decisions about making payouts to investors and employees.\n>\n> *[Corporate structure — image]* \"Another provision from our paperwork specifies that the nonprofit retains control. (The paperwork uses OpenAI LP's official name 'OpenAI, L.P.'.)\"\n>\n> As mentioned above, economic returns for investors and employees are capped (with the cap negotiated in advance on a per-limited partner basis). Any excess returns are owned by the nonprofit. Our goal is to ensure that most of the value we create if successful is returned to the world, so we think this is an important first step. **Returns for our first round of investors are capped to 100x their investment**, and we expect this multiple to be lower for future rounds.\n>\n> **What OpenAI does**\n>\n> Our day-to-day work remains the same. Today, we believe we can build the most value by focusing exclusively on developing new AI technologies, not commercial products. Our structure gives us flexibility for how to make money in the long term, but we hope to figure that out only once we've created safe AGI (though we're open to non-distracting revenue sources such as licensing in the interim).\n>\n> OpenAI LP currently employs around 100 people organized into three main areas: capabilities (advancing what AI systems can do), safety (ensuring those systems are aligned with human values), and policy (ensuring appropriate governance for such systems). OpenAI Nonprofit governs OpenAI LP, runs educational programs such as Scholars and Fellows, and hosts policy initiatives. OpenAI LP is continuing (at increased pace and scale) the development roadmap started at OpenAI Nonprofit, which has yielded breakthroughs in reinforcement learning, robotics, and language.\n>\n> **Safety**\n>\n> We are concerned about AGI's potential to cause rapid change, whether through machines pursuing goals misspecified by their operator, malicious humans subverting deployed systems, or an out-of-control economy that grows without resulting in improvements to human lives. As described in our Charter, we are willing to merge with a value-aligned organization (even if it means reduced or zero payouts to investors) to avoid a competitive race which would make it hard to prioritize safety.\n>\n> **Who's involved**\n>\n> * OpenAI Nonprofit's board consists of OpenAI LP employees Greg Brockman (Chairman & CTO), Ilya Sutskever (Chief Scientist), and Sam Altman (CEO), and non-employees Adam D'Angelo, Holden Karnofsky, Reid Hoffman, Sue Yoon, and Tasha McCauley.\n> * **Elon Musk left the board of OpenAI Nonprofit in February 2018 and is not involved with OpenAI LP.**\n> * Our investors include Reid Hoffman and Khosla Ventures.\n>\n> We are traveling a hard and uncertain path, but we have designed our structure to help us positively affect the world should we succeed in creating AGI. If you'd like to help us make this mission a reality, we're hiring :)!\n\n## Commentary\n\nPX 239 is plaintiffs' clean documentary anchor for two propositions: (i) **Altman walked Musk through the capped-profit structure in advance and named the \"multi-billion dollar investment\" being negotiated** (the Microsoft Series A, which closed July 22, 2019 at $1B and $20B target redemption — see [[Key Themes]] §\"Day 7 — the Microsoft economics\"); and (ii) the post itself **publicly committed** OpenAI LP to \"primary fiduciary obligation is to advance the aims of the OpenAI Charter\" and \"the company is controlled by OpenAI Nonprofit's board.\" The \"**purple box**\" image referenced here is the same box that becomes [[DX-827]] (the LP Summary Term Sheet) and the centerpiece of defense's \"Musk knew, Musk consented\" cross — Birchall's \"infamous purple box\" exchange (see [[Key Themes]]). Plaintiffs read the email as Musk's pre-publication notice and consent to a structure Altman represented would put mission first; defense reads it as Musk-was-told-everything-and-said-nothing. The **\"100x cap\"** in the post and the post's silence on Microsoft are both relevant to plaintiffs' SOL response — Musk's Sept 2020 \"captured by Microsoft\" tweet ([[Key Themes]]) followed less than 18 months later.\n\n---\n*See also:* [[Day 3|Day 3 digest]] · [[DX-827]] · [[Sam Altman]] · [[Greg Brockman]] · [[Shivon Zilis]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-24", "exhibit": "PX 24", "party": "Plaintiffs", "type": "Charter (web page)", "admitted_trial_day": "Day 4 (April 30, 2026)", "uploaded_box_pt": "2026-04-30T15:26:47", "uploader": "Someone", "pages": 5, "size_bytes": 714243, "source_pdf": "PX-24.pdf", "pdf_url": "https://media.mts-in.com/PX-24.pdf", "body_markdown": "# PX 24 — OpenAI Charter (April 9, 2018)\n\n> The OpenAI Charter as published — the document plaintiffs cite for the proposition that OpenAI's \"primary fiduciary duty is to humanity.\"\n\n## Document type\n**Charter / public website printout.** A 5-page printout of `production.openai.com/charter` captured June 3, 2025, showing the published version of OpenAI's April 9, 2018 Charter — the document the 2018 for-profit conversion was framed around. Bates OPENAI_MUSK00012044–048.\n\n## Logistics\n- **Trial admission:** Day 4 (April 30, 2026) — used during Birchall direct/cross to anchor the charitable-trust framing alongside the LP term sheet ([[DX 827]]). No prior wiki reference; admission day inferred from Box upload date.\n- **Box upload:** 2026-04-30 15:26:47 PT — late-afternoon Day 4 batch (clustered with PX 71, 87, 103, 1504, DX 539, 646, 1156, 1285).\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~714 KB, 5 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `024.pdf`.\n\n## Transcribed text\n\n> **OpenAI Charter**\n> Our Charter describes the principles we use to execute on OpenAI's mission.\n>\n> **Published April 9, 2018**\n>\n> This document reflects the strategy we've refined over the past two years, including feedback from many people internal and external to OpenAI. The timeline to AGI remains uncertain, but our Charter will guide us in acting in the best interests of humanity throughout its development.\n>\n> OpenAI's mission is to ensure that artificial general intelligence (AGI)—by which we mean highly autonomous systems that outperform humans at most economically valuable work—benefits all of humanity. We will attempt to directly build safe and beneficial AGI, but will also consider our mission fulfilled if our work aids others to achieve this outcome. To that end, we commit to the following principles:\n>\n> **Broadly distributed benefits**\n>\n> We commit to use any influence we obtain over AGI's deployment to ensure it is used for the benefit of all, and to avoid enabling uses of AI or AGI that harm humanity or unduly concentrate power.\n>\n> Our primary fiduciary duty is to humanity. We anticipate needing to marshal substantial resources to fulfill our mission, but will always diligently act to minimize conflicts of interest among our employees and stakeholders that could compromise broad benefit.\n>\n> **Long-term safety**\n>\n> We are committed to doing the research required to make AGI safe, and to driving the broad adoption of such research across the AI community.\n>\n> We are concerned about late-stage AGI development becoming a competitive race without time for adequate safety precautions. Therefore, if a value-aligned, safety-conscious project comes close to building AGI before we do, we commit to stop competing with and start assisting this project. We will work out specifics in case-by-case agreements, but a typical triggering condition might be \"a better-than-even chance of success in the next two years.\"\n>\n> **Technical leadership**\n>\n> To be effective at addressing AGI's impact on society, OpenAI must be on the cutting edge of AI capabilities—policy and safety advocacy alone would be insufficient.\n>\n> We believe that AI will have broad societal impact before AGI, and we'll strive to lead in those areas that are directly aligned with our mission and expertise.\n>\n> **Cooperative orientation**\n>\n> We will actively cooperate with other research and policy institutions; we seek to create a global community working together to address AGI's global challenges.\n>\n> We are committed to providing public goods that help society navigate the path to AGI. Today this includes publishing most of our AI research, but we expect that safety and security concerns will reduce our traditional publishing in the future, while increasing the importance of sharing safety, policy, and standards research.\n\n[Remainder: pages 4–5 — site footer/navigation links only.]\n\n## Commentary\n\nThe Charter is the central public artifact of OpenAI's 2018 reorganization and the textual foundation for plaintiffs' charitable-trust theory. Two phrases do most of the work: \"Our **primary fiduciary duty is to humanity**\" and the \"**stop competing with and start assisting**\" clause (the so-called merge-and-assist provision). Both reappear repeatedly across the Brockman journal — see Brockman's January 31, 2018 letter \"Our biggest tool is the moral high ground… our fiduciary duty should be to humanity\" ([[DX 748]]) and the September 12, 2017 entry quoting Musk (\"Must tell the story and not lose moral high ground\") in [[PX 154]]. Plaintiffs' Day 7 cross of Robert Wu (the OpenAI 30(b)(6) witness) used the Charter as the baseline against which to measure the December 2023 unpublished \"Charter 2.0\" rewrite — which removed the \"primary fiduciary duty is to humanity\" language and added \"we've grown to regard capitalism not as a constraint, but instead as a positive force.\" See [[Key Themes]] § \"Charter 2.0\".\n\n---\n*See also:* [[Day 4|Day 4 digest]] · [[Day 7|Day 7 digest]] · [[Brockman Journal]] · [[Key Themes]] · [[DX 827]] · [[DX 862]]\n"}
{"exhibit_id": "PX-241", "exhibit": "PX 241", "party": "Plaintiffs", "type": "Blog post", "admitted_trial_day": "Day 6 (May 4, 2026)", "uploaded_box_pt": "2026-05-04T14:46:12", "uploader": "Someone", "pages": 6, "size_bytes": 6257102, "source_pdf": "PX-241.pdf", "pdf_url": "https://media.mts-in.com/PX-241.pdf", "body_markdown": "# PX 241 — March 11, 2019 \"OpenAI LP\" announcement blog post\n\n> The six-page printout of OpenAI's March 11, 2019 announcement of OpenAI LP — the \"capped-profit\" hybrid that the founders launched after the August–November 2017 negotiation Brockman recorded in [[PX-151]] and [[PX-161]]. The post coins \"capped-profit,\" reproduces the **purple-box** disclaimer language, names the new board (Brockman, Sutskever, Altman + D'Angelo, Karnofsky, Hoffman, Zilis, McCauley), and announces \"**Elon Musk left the board of OpenAI Nonprofit in February 2018 and is not formally involved with OpenAI LP.**\"\n\n## Document type\n**Blog post (printout), six pages.** Source: openai.com/index/openai-lp/. Authors: Greg Brockman, Ilya Sutskever, OpenAI. Bates 2024MUSK-0014452–0014457. Includes the Turner-style watercolor banner image and the in-line \"purple box\" disclaimer language reproduced verbatim from OpenAI LP investor paperwork.\n\n## Logistics\n- **Trial admission:** Day 6 (May 4, 2026). Used by Mr. Kry during Brockman cross to set up the textual contrast: the nonprofit-board-controls / \"capped-profit\" framing in the announcement vs. the journal entries in [[PX-151]] and [[PX-161]] showing the founders' contemporaneous private preference for B-corp economics.\n- **Box upload:** 2026-05-04 14:46:12 PT — Day 6 mid-afternoon batch (clustered with PX 151, PX 161, PX 390).\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~6.0 MB, 6 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `241.pdf`.\n\n## Transcribed text\n\n> **March 11, 2019 — Company**\n> **OpenAI LP**\n>\n> Our mission is to ensure that artificial general intelligence (AGI) benefits all of humanity, primarily by attempting to build safe AGI and share the benefits with the world.\n>\n> We've experienced firsthand that the most dramatic AI systems use the most computational power in addition to algorithmic innovations, and decided to scale much faster than we'd planned when starting OpenAI. We'll need to invest billions of dollars in upcoming years into large-scale cloud compute, attracting and retaining talented people, and building AI supercomputers.\n>\n> We want to increase our ability to raise capital while still serving our mission, and no pre-existing legal structure we know of strikes the right balance. Our solution is to create OpenAI LP as a hybrid of a for-profit and nonprofit—which we are calling a \"capped-profit\" company.\n>\n> The fundamental idea of OpenAI LP is that investors and employees can get a capped return if we succeed at our mission, which allows us to raise investment capital and attract employees with startup-like equity. But any returns beyond that amount—and if we are successful, we expect to generate orders of magnitude more value than we'd owe to people who invest in or work at OpenAI LP—are owned by the original OpenAI Nonprofit entity.\n>\n> Going forward (in this post and elsewhere), \"OpenAI\" refers to OpenAI LP (which now employs most of our staff), and the original entity is referred to as \"OpenAI Nonprofit.\"\n>\n> ## The mission comes first\n>\n> We've designed OpenAI LP to put our overall mission—ensuring the creation and adoption of safe and beneficial AGI—ahead of generating returns for investors.\n>\n> The mission comes first even with respect to OpenAI LP's structure. While we are hopeful that what we describe below will work until our mission is complete, we may update our implementation as the world changes. Regardless of how the world evolves, we are committed—legally and personally—to our mission.\n>\n> OpenAI LP's primary fiduciary obligation is to advance the aims of the OpenAI Charter, and the company is controlled by OpenAI Nonprofit's board. All investors and employees sign agreements that OpenAI LP's obligation to the Charter always comes first, even at the expense of some or all of their financial stake.\n>\n> > **IMPORTANT**\n> >\n> > **The Partnership exists to advance OpenAI Inc's mission of ensuring that safe artificial general intelligence is developed and benefits all of humanity. The General Partner's duty to this mission and the principles advanced in the OpenAI Inc Charter take precedence over any obligation to generate a profit. The Partnership may never make a profit, and the General Partner is under no obligation to do so. The General Partner is free to re-invest any or all of the Operating Entity's (or the Partnership's) cash flow into research and development activities and/or related expenses without any obligation to the Limited Partners. See Section 6.4 of the Operating Entity's Limited Partnership Agreement for additional details.**\n>\n> Our employee and investor paperwork start with big purple boxes like this. The general partner refers to OpenAI Nonprofit (whose legal name is \"OpenAI Inc\"); limited partners refers to investors and employees.\n>\n> Only a minority of board members are allowed to hold financial stakes in the partnership at one time. Furthermore, only board members without such stakes can vote on decisions where the interests of limited partners and OpenAI Nonprofit's mission may conflict—including any decisions about making payouts to investors and employees.\n>\n> > **Corporate Structure** — OpenAI, L.P. will be a for-profit Delaware Limited Partnership managed by its General Partner, a single-member Delaware LLC controlled by OpenAI, Inc. (the Nonprofit)'s Board of Directors. At all times, no more than a minority of the Nonprofit's Board of Directors will be holders of any economic interest in OpenAI, L.P. or in the employee holdings entity (described below).\n>\n> Another provision from our paperwork specifies that OpenAI Nonprofit retains control.\n>\n> As mentioned above, economic returns for investors and employees are capped (with the cap negotiated in advance on a per-limited partner basis). Any excess returns go to OpenAI Nonprofit. Our goal is to ensure that most of the value (monetary or otherwise) we create if successful benefits everyone, so we think this is an important first step. **Returns for our first round of investors are capped at 100x their investment** (commensurate with the risks in front of us), and we expect this multiple to be lower for future rounds as we make further progress.\n>\n> ## What OpenAI does\n>\n> Our day-to-day work is not changing. Today, we believe we can build the most value by focusing exclusively on developing new AI technologies, not commercial products. Our structure gives us flexibility for how to create a return in the long term, but we hope to figure that out only once we've created safe AGI.\n>\n> OpenAI LP currently employs around 100 people organized into three main areas: capabilities (advancing what AI systems can do), safety (ensuring those systems are aligned with human values), and policy (ensuring appropriate governance for such systems). OpenAI Nonprofit governs OpenAI LP, runs educational programs such as Scholars and Fellows, and hosts policy initiatives. OpenAI LP is continuing (at increased pace and scale) the development roadmap started at OpenAI Nonprofit, which has yielded breakthroughs in reinforcement learning, robotics, and language.\n>\n> ## Safety\n>\n> We are excited by the potential for AGI to help solve planetary-scale problems in areas where humanity is failing and there is no obvious solution today. However, we are also concerned about AGI's potential to cause rapid change, whether through machines pursuing goals misspecified by their operator, malicious humans subverting deployed systems, or an out-of-control economy that grows without resulting in improvements to human lives. As described in our Charter, we are willing to merge with a value-aligned organization (even if it means reduced or zero payouts to investors) to avoid a competitive race which would make it hard to prioritize safety.\n>\n> ## Who's involved\n>\n> - OpenAI Nonprofit's board consists of OpenAI LP employees Greg Brockman (Chairman & CTO), Ilya Sutskever (Chief Scientist), and Sam Altman (CEO), and non-employees Adam D'Angelo, Holden Karnofsky, Reid Hoffman, Shivon Zilis, and Tasha McCauley.\n> - **Elon Musk left the board of OpenAI Nonprofit in February 2018 and is not formally involved with OpenAI LP. We are thankful for all his past help.**\n> - Our investors include Reid Hoffman's charitable foundation and Khosla Ventures, among others. We feel lucky to have mission-aligned, impact-focused, helpful investors!\n>\n> *We are traveling a hard and uncertain path, but we have designed our structure to help us positively affect the world should we succeed in creating AGI—which we think will have as broad impact as the computer itself and improve healthcare, education, scientific research, and many aspects of people's lives. If you'd like to help us make this mission a reality, we're hiring :)!*\n>\n> **Authors:** Greg Brockman, Ilya Sutskever, OpenAI · 2019\n>\n> [Footnote A: This list represents the board as of March 11, 2019 and does not reflect further board member updates.]\n\n## Commentary\n\nPX 241 is the public face of what [[PX-151]] and [[PX-161]] document privately. The post is plaintiffs' \"morally-high-ground\" Exhibit A: in March 2019, OpenAI told the world the new entity's \"primary fiduciary obligation is to advance the aims of the OpenAI Charter,\" that \"the mission comes first,\" that the nonprofit retains voting control, that the **purple-box** disclaimer warns investors \"the Partnership may never make a profit,\" and that returns are capped at 100×. Plaintiffs use it as the textual benchmark against which the **2025 Watershed MOU** (Robert Wu's Day 7 testimony) and the **September 2024 \"uncapped stock\" plan** (Wu, \"converting the LLC waterfall into a more traditional corporate stock ownership capital structure\") are measured — see [[Key Themes]] § \"Day 7 — the Microsoft economics.\" Defense reads PX 241 as the receipt for **mission-first structuring**: the nonprofit retained control, the cap was real for ~$1B-vintage investors, and Musk had already left voluntarily in February 2018, more than a year before the LP launch. The \"**Elon Musk left the board of OpenAI Nonprofit in February 2018**\" sentence — **\"We are thankful for all his past help\"** — is in evidence as Musk's own framing-point on cross: defense argues this is the public moment Musk had constructive notice of the LP structure, triggering the statute of limitations.\n\n---\n*See also:* [[Day 6|Day 6 digest]] · [[Day 7|Day 7 digest]] · [[PX-24]] · [[PX-1504]] · [[DX-827]] · [[Watershed MOU]] · [[Statute of Limitations]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-244", "exhibit": "PX 244", "party": "Plaintiffs", "type": "Email", "admitted_trial_day": "Day 8 (May 6, 2026)", "uploaded_box_pt": "2026-05-06T14:07:56", "uploader": "Someone", "pages": 1, "size_bytes": 197168, "source_pdf": "PX-244.pdf", "pdf_url": "https://media.mts-in.com/PX-244.pdf", "body_markdown": "# PX 244 — May 11, 2019 Mira Murati → Phil Waymouth (Microsoft) — \"OpenAI Edits\" to the Odyssey JDCA\n\n> Murati's transmittal of OpenAI's redlines to the May 10, 2019 Microsoft \"Odyssey Joint Dev and Collab Agreement\" — including her live-negotiation note that the **open-source posture on language models** \"continues to be sensitive\" and her offer of a **6-month-or-1-year exclusivity window** to Microsoft \"prior to open sourcing what may be considered a better version of the technology.\"\n\n## Document type\n**Email**, plain text, single message. Murati (mira@openai.com) → Phil Waymouth (Microsoft); attaching `MSFT - Odyssey Joint Dev and Collab Agr 5-10-19 (OpenAI Edits).docx`. Bates MSFT_MUSK000015915. Marked HIGHLY CONFIDENTIAL.\n\n## Logistics\n- **Trial admission:** Day 8 (May 6, 2026) — used during Mira Murati's live testimony. The Odyssey agreement is the antecedent of what becomes the 2023 JDCA referenced in [[Key Themes]] §\"Day 7 — the Microsoft economics\" (\"OpenAI Inc., OpCo, licensed [Microsoft] all IP and technology … excluding only AGI\").\n- **Box upload:** 2026-05-06 14:07:56 PT — Day 8 mid-afternoon Plaintiffs' batch.\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~193 KB, 1 page.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `244.pdf`.\n\n## Transcribed text\n\n> **From:** Mira Murati <mira@openai.com>\n> **To:** Phil Waymouth <[redacted]@microsoft.com>\n> **Subject:** OpenAI Edits\n> **Date:** 5/11/2019 7:00:32 AM\n> **Attachments:** MSFT - Odyssey Joint Dev and Collab Agr 5-10-19 (OpenAI Edits).docx\n>\n> Hi Phil,\n>\n> Thank you very much for making the time to discuss the agreement in person today. Hope you had a smooth flight back.\n>\n> I apologize for taking longer than expected in getting the digital copy of the redlines back to you. Everything should reflect what we discussed, **with the exception of the open source issue under the language model**. I discussed it more with the team and **this continues to be sensitive because we plan to open source the current version and it is not unreasonable that we may continue to open source future versions of the language model**. However, I am wondering if we could provide you with say **6 months or 1 year of exclusivity prior to open sourcing what may be considered a better version of the technology**. And if we choose to commercially license future versions then of course you would have first right of negotiation anyway. Let me know your thoughts.\n>\n> Please see attached the document and I am happy to discuss over the phone during the weekend, if helpful.\n>\n> Thank you!\n> --\n> Mira\n\n## Commentary\n\nPX 244 is plaintiffs' clean exhibit on the **mid-2019 Microsoft-OpenAI commercial-IP negotiation** — exactly the period in which Musk says (and the Sept 2020 \"captured by Microsoft\" tweet asserts) that the for-profit \"captured\" the mission. Murati's note that OpenAI \"**plans to open source the current version**\" of the language model and that \"it is not unreasonable that we may continue to open source future versions\" is plaintiffs' contemporaneous corroboration of Musk's \"OPEN\" / open-source theme on the stand (see [[Key Themes]] §\"The 'OPEN' in OpenAI\"). At the same time, the proposed **6-month/1-year exclusivity window** for Microsoft and Microsoft's \"first right of negotiation\" on commercial licensing are the seed of the architecture that culminates in the 2023 JDCA \"all IP and technology … excluding only AGI\" carveout (see [[Key Themes]] §\"The 2023 JDCA\"). The exhibit was used on Day 8 to bracket Murati's later, more guarded testimony about Altman's candor and the Microsoft pressure during the November 2023 firing — see [[PX 306]] for the Murati ↔ Nadella iMessage thread from that week.\n\n---\n*See also:* [[Day 8|Day 8 digest]] · [[PX 306]] · [[Mira Murati]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-25", "exhibit": "PX 25", "party": "Plaintiffs", "type": "Email thread", "admitted_trial_day": "Day 3 (April 29, 2026)", "uploaded_box_pt": "2026-04-29T15:09:19", "uploader": "Someone", "pages": 1, "size_bytes": 209187, "source_pdf": "PX-25.pdf", "pdf_url": "https://media.mts-in.com/PX-25.pdf", "body_markdown": "# PX 25 — Nov 20, 2015 Musk → Altman \"Re: AI docs\" (Delaware non-profit + parallel structure)\n\n> Musk's terse one-paragraph reply pushing back on Altman's proposed YC-subsidiary structure for \"YC AI\" — telling Altman to keep the new venture independent of YC and recommending \"a standard C corp with a parallel nonprofit.\"\n\n## Document type\n**Email thread, plain text, two messages.** Musk's November 20, 2015 8:29 PM (UTC) reply on top of Altman's same-day 11:48 AM original. Subject \"Re: AI docs.\" Recipients Altman → Musk only. Bates 2024MUSK-0009963.\n\n## Logistics\n- **Trial admission:** Day 3 (April 29, 2026). No prior wiki reference; admission day inferred from Box upload and Day 3 mid-afternoon plaintiffs' batch (clustered with PX 105, PX 158, PX 236).\n- **Box upload:** 2026-04-29 15:09:19 PT.\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~204 KB, 1 page.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `025.pdf`.\n\n## Transcribed text\n\n> **From:** \"Elon Musk\" <erm@spacex.com>\n> **To:** \"Sam Altman\" <sam@ycombinator.com>\n> **Subject:** Re: AI docs\n> **Date:** Fri, 20 Nov 2015 20:29:18 -0000\n> **Importance:** Normal\n>\n> I think this should be independent from (but supported by) YC, not what sounds like a subsidiary.\n>\n> Also, the structure doesn't seem optimal. In particular, the YC stock along with a salary from the nonprofit muddies the alignment of incentives. Probably better to have a standard C corp with a parallel nonprofit.\n>\n> ---\n>\n> On Nov 20, 2015, at 11:48 AM, Sam Altman <sam@ycombinator.com> wrote:\n>\n> > Elon--\n> >\n> > Plan is to have you, me, and Ilya on the Board of Directors for YC AI, which will be a Delaware non-profit. We will also state that we plan to elect two other outsiders by majority vote of the Board.\n> >\n> > We will write into the bylaws that any technology that potentially compromises the safety of humanity has to get consent of the Board to be released, and we will reference this in the researchers' employment contracts.\n> >\n> > At a high level, does that work for you?\n> >\n> > I'm cc'ing our GC Jon Levy here--is there someone in your office he can work with on the details?\n> >\n> > Sam\n\n## Commentary\n\nPX 25 captures the November 20, 2015 hour in which the founding structure of OpenAI was decided — three weeks before launch. Plaintiffs use this email two ways: (i) Altman's bylaws commitment that **\"any technology that potentially compromises the safety of humanity has to get consent of the Board to be released\"** — a written safety-veto promise the plaintiffs frame as part of the charitable-trust bargain; and (ii) Musk's own response, *\"Probably better to have a standard C corp with a parallel nonprofit,\"* which the defense reads as Musk himself contemplating a for-profit/nonprofit hybrid from the start (compatible with the eventual 2019 LP structure). The \"independent from (but supported by) YC\" line cuts against later defense suggestions that Musk understood OpenAI as a YC-controlled effort. Cross-reference: this email is the proximate precursor to the December 8, 2015 articles of incorporation ([[PX 5]]) and the December 11, 2015 OpenAI launch announcement.\n\n---\n*See also:* [[Day 3|Day 3 digest]] · [[Sam Altman]] · [[PX 5]] · [[PX 7]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-251", "exhibit": "PX 251", "party": "Plaintiffs", "type": "Tweet thread", "admitted_trial_day": "Day 3 (April 29, 2026)", "uploaded_box_pt": "2026-04-29T15:09:18", "uploader": "Someone", "pages": 1, "size_bytes": 312520, "source_pdf": "PX-251.pdf", "pdf_url": "https://media.mts-in.com/PX-251.pdf", "body_markdown": "# PX 251 — Sept 24, 2020 Musk tweet, \"OpenAI is essentially captured by Microsoft\"\n\n> A one-page screenshot of a three-tweet X/Twitter thread on September 24, 2020 — Whole Mars Catalog and \"Disrupt or Distort\" tweeting about Microsoft's exclusive license to OpenAI's GPT-3, and Musk replying: \"**This does seem like the opposite of open. OpenAI is essentially captured by Microsoft.**\"\n\n## Document type\n**Tweet thread, screenshot.** Three-tweet image — Whole Mars Catalog (@WholeMarsBlog) Sep 24, 2020 quote-tweeting a VentureBeat headline (\"Microsoft gets exclusive license for OpenAI's GPT-3 language model\") and tagging @elonmusk, \"Disrupt or Distort\" (@asymmetricfocus) replying (\"'Exclusive License'. Might as well rename it to 'ClosedAI'\"), and **Elon Musk (@elonmusk) at 4:50 AM · Sep 24, 2020**: \"This does seem like the opposite of open. OpenAI is essentially captured by Microsoft.\" Engagement: 149 replies / 264 reposts / 1.9K likes / 38 bookmarks. Bates MSFT_MUSK000085666.\n\n## Logistics\n- **Trial admission:** Day 3 (April 29, 2026 — Musk direct concludes; cross begins). Per the Day 3 transcript table: \"PX 251 | Sept. 24, 2020 Musk Twitter post — 'OpenAI is essentially captured by Microsoft'\" `(042926TT.txt:2154–2179)`. Re-used Day 4 cross of Musk `(043026TT.txt:2278–2295)`.\n- **Box upload:** 2026-04-29 15:09:18 PT — Day 3 mid-afternoon batch (clustered with PX 98, PX 157, PX 233, PX 296, PX 355).\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~305 KB, 1 page.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `251.pdf`. Bates is `MSFT_MUSK000085666` — i.e., Microsoft produced the screenshot, not Musk.\n\n## Transcribed text\n\n> **Whole Mars Catalog ✓** @WholeMarsBlog · Sep 24, 2020\n> \"Microsoft gets exclusive license for OpenAI's GPT-3 language model\"\n> I thought OpenAI was supposed to democratize this tech… not give Microsoft an exclusive license. @elonmusk\n> [embedded VentureBeat link card with the OpenAI logo and the headline]\n> 35 / 90 / 570\n>\n> **Disrupt or Distort ✓** @asymmetricfocus · Sep 24, 2020\n> \"Exclusive License\". Might as well rename it to \"ClosedAI\".\n> 6 / 16 / 497\n>\n> **Elon Musk ✓ 𝕏** @elonmusk\n> This does seem like the opposite of open. OpenAI is essentially captured by Microsoft.\n> 4:50 AM · Sep 24, 2020\n> 149 / 264 / 1.9K / 38\n\n## Commentary\n\nPX 251 is the **single most important exhibit on the statute-of-limitations fight**. Defense — Cohen for Microsoft especially — uses it to argue that Musk's claims accrued in **September 2020**: by his own public admission, Musk understood OpenAI to be \"**captured by Microsoft**\" four years before he sued, and a four-year-old breach is outside the SOL window. Plaintiffs' counter, drawn from Musk's own Day 3 testimony at 042926TT.txt:2194, is that \"**Sam Altman immediately reached out to reassure me that OpenAI was on its — was staying on mission as a nonprofit**\" — i.e., the tweet was retracted by reassurance, the breach was concealed, and the \"[[Key Themes|three phases]]\" framework places September 2020 still in *phase two* (uncertain, not yet resolved into \"looting\"). The fact that the screenshot's Bates is **Microsoft's** production, not Musk's, is itself a defense beat: Microsoft preserved the tweet for the cross of its own customer-adverse witness. The tweet is also the **textual link** between the [[Key Themes|\"OPEN\" in OpenAI]] theme (Musk: \"the opposite of open\") and the Microsoft-capture theme that runs through Day 7 (the **2023 JDCA** licensing of \"all IP and technology … excluding only AGI\" — see [[Key Themes]] § \"Day 7 — Microsoft economics\"). See also Birchall on cross, Day 4, who handled the press response that followed.\n\n---\n*See also:* [[Day 3|Day 3 digest]] · [[Day 4|Day 4 digest]] · [[Statute of Limitations]] · [[Microsoft]] · [[Sam Altman]] · [[Elon Musk]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-252", "exhibit": "PX 252", "party": "Plaintiffs", "type": "Text messages", "admitted_trial_day": "Day 3 (April 29, 2026)", "uploaded_box_pt": "2026-04-29T15:09:18", "uploader": "Someone", "pages": 2, "size_bytes": 72347, "source_pdf": "PX-252.pdf", "pdf_url": "https://media.mts-in.com/PX-252.pdf", "body_markdown": "# PX 252 — Oct 28, 2020 Altman text to Musk re \"the next Microsoft investment\"\n\n> Two-message Cellebrite extract: Sam Altman texts Musk asking for advice on \"the next Microsoft investment\" the same week as Musk's [[PX-251|\"OpenAI is essentially captured by Microsoft\"]] tweet — plaintiffs' contemporaneous corroboration that Altman *actively reached out* to Musk in 2020.\n\n## Document type\n**Text messages (forensic extract).** Cellebrite-style \"Short Message Report,\" 1 conversation, 5 listed participants, 2 messages, date range 10/28/2020. Sender identifiers include \"Sam Altman\" and \"Elon Musk <erm@tesla.com>.\" Bates 2024MUSK-0006317 through 2024MUSK-0006318.\n\n## Logistics\n- **Trial admission:** Day 3 (April 29, 2026) — Musk direct concludes; cross begins.\n- **Box upload:** 2026-04-29 15:09:18 PT — Day 3 mid-afternoon batch (clustered with PX 99, PX 157, PX 233, PX 296).\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~71 KB, 2 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `252.pdf`.\n\n## Transcribed text\n\n> **Short Message Report**\n> Conversations: 1 / Participants: 5 / Total Messages: 2 / Date Range: 10/28/2020\n>\n> 2 messages on 10/28/2020 — Elon Musk · Elon Musk <erm@tesla.com> · Sam Altman\n\n**Messages in chronological order (times shown in GMT -07:00):**\n\n> **SA — Sam Altman** — 10/28/2020, 11:37 AM\n> i would love to get you some advice from you on the next microsoft investment we are thinking of if you have time in the next week or so\n>\n> *Receipts · Elon Musk [R: 10/28/2020, 2:52 PM]*\n>\n> **EM — Elon Musk** — 10/28/2020, 2:53 PM\n> Ok. I can talk probably tomorrow or the next day.\n>\n> *Receipts · Sam Altman [D: 10/28/2020, 2:53 PM]*\n\n## Commentary\n\nPX 252 lands directly on the [[Key Themes|\"captured by Microsoft\" / statute-of-limitations fight]]. Musk's [[PX-251|Sept 24, 2020 tweet]] — \"**This does seem like the opposite of open. OpenAI is essentially captured by Microsoft**\" — was just five weeks earlier. Musk's testimony explaining why he didn't sue then was that \"**Sam Altman immediately reached out to reassure me that OpenAI was on its — was staying on mission as a nonprofit.**\" PX 252 is documentary corroboration of exactly that pattern of reach-out — Altman, on Oct 28, 2020, asking Musk for \"advice\" on \"the next microsoft investment we are thinking of,\" and Musk agreeing to talk within 24 hours. **Plaintiffs use it for SOL** (\"Altman kept reaching out to keep me on the hook\"). **Defense's counter-read** is the opposite: a 2020 Musk who is in two-way text contact with Altman about Microsoft round mechanics is on actual notice of the relevant facts, and the breach (if any) became actually visible then — which would moot the limitations argument from the other direction. See [[Day 3|Day 3 digest]] for Musk's direct testimony on this period and [[Key Themes]] for the broader Microsoft-thread arc.\n\n---\n*See also:* [[Day 3|Day 3 digest]] · [[PX-251]] · [[Sam Altman]] · [[Microsoft Thread]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-284", "exhibit": "PX 284", "party": "Plaintiffs", "type": "Court filing", "admitted_trial_day": "Day 8 (May 6, 2026)", "pages": 6, "size_bytes": 312298, "source_pdf": "PX-284.pdf", "bates": "284.001-284.003", "pdf_url": "https://media.mts-in.com/PX-284.pdf", "body_markdown": "# PX 284 — Robert Wu Supplemental Declaration (October 17, 2025)\n\n> The **supplemental declaration of Robert Wu**, OpenAI's deputy general counsel and Rule 30(b)(6) corporate representative, attesting to OpenAI's TRA framework and the OpenAI Inc. Board's role in approving each TRA. Two paragraphs of this declaration were **read into the trial record on Day 8 (May 6, 2026)** under a limiting instruction admitting the evidence **only against OpenAI, not Microsoft**.\n\n## Document type\n**Court filing — declaration under penalty of perjury** (FRE 1101 / 28 U.S.C. § 1746). Sworn before Morrison & Foerster + Wachtell, Lipton lawyers; Docusign Envelope ID `D99047E6-F5BB-4CB5-BEF6-AA8D8DFB01C5`. Signed October 17, 2025 in San Francisco.\n\n## Logistics\n- **Trial admission:** **Day 8 (May 6, 2026).** Read into the record by **Mr. Einen** under a limiting instruction from the Court (\"**It cannot be offered against Microsoft. … this is only as against OpenAI, not Microsoft.**\" *(5/6/2026 Testimony @ ~9:02 PT)*).\n- **File size:** ~305 KB, 6 pages (cover sheet + 3-page declaration with Docusign signature page).\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `284.pdf`.\n- **Counsel:** Jordan Eth, William Frentzen, David J. Wiener (MoFo); William Savitt, Bradley R. Wilson, Sarah K. Eddy, Steven Winter, Nathaniel Cullerton (Wachtell, Lipton).\n\n## Transcribed text\n\n> **DECLARATION OF ROBERT WU**\n>\n> I, ROBERT WU, declare as follows:\n>\n> **1.** I serve as Deputy General Counsel for OpenAI OpCo, LLC, a subsidiary of OpenAI, Inc. (together with its affiliates, \"OpenAI\"). On October 3, 2025, I testified on behalf of OpenAI, Inc. as its corporate representative pursuant to Federal Rule of Civil Procedure 30(b)(6), and it is in that capacity that I provide this Declaration.\n>\n> **2.** Between 2019 and 2023, OpenAI's employees and certain third-party investors received economic interests in OpenAI's for-profit subsidiary that are subject to **target redemption amounts, or \"TRAs,\" that establish the maximum possible return on each investment**. These TRAs were often but not always based on a multiple of the amount of invested capital. **For example, in 2019, Microsoft Corporation invested approximately $1 billion based on a 20x multiple, resulting in a TRA of approximately $20 billion.**\n>\n> **3.** OpenAI's management team was responsible for negotiating the terms of these economic interests in OpenAI's for-profit subsidiary, including the TRAs. **These negotiations occurred at arm's length and were subject to oversight and ultimate approval by the OpenAI, Inc. Board of Directors.**\n>\n> **4.** The OpenAI, Inc. Board was responsible for reviewing the proposed terms of each potential economic interest in OpenAI's for-profit subsidiary, including the applicable TRAs, and determining whether those terms were in the best interests of OpenAI, Inc. When making this determination, the Board considered a variety of factors, including the extent to which additional capital was necessary for OpenAI, Inc. to accomplish its mission; **the Board's desire to maintain, to the greatest extent possible, the value of OpenAI, Inc.'s residual economic interest in OpenAI's for-profit subsidiary**; and information regarding the terms on which potential investors in OpenAI's for-profit subsidiary would be willing to invest at the relevant time.\n>\n> **5.** When negotiating and evaluating the TRAs and other proposed terms of these economic interests in OpenAI's for-profit subsidiary, **the OpenAI, Inc. Board and OpenAI management team had access to the periodic valuation reports that OpenAI received from PricewaterhouseCoopers and Andersen Consulting in the ordinary course of OpenAI's business**. These reports used multiple methodologies to estimate the fair value of the equity and other economic interests in OpenAI's for-profit subsidiary, taking into account, among other things, (a) the OpenAI management team's forecasts of future business operations; and (b) the terms of any economic interests in the for-profit subsidiary that existed on the relevant valuation date, including the applicable TRAs and the mechanics of the profit distribution waterfall. **It is my understanding that these periodic valuation reports have been produced in this litigation.**\n>\n> **6.** The OpenAI, Inc. Board did not consider any financial or quantitative analysis specifically prepared by a third-party financial advisor to assess any proposed TRAs.\n>\n> I declare under penalty of perjury that the foregoing is true and correct to the best of my knowledge.\n>\n> Executed on October 17, 2025, at San Francisco, California.\n>\n> /s/ **Robert Wu**\n\n## Commentary\n\nPX 284 is the document underneath the Day 8 morning's TRA framework testimony. Its function in plaintiffs' case is to **authenticate three structural concessions** without putting Wu back on a witness stand:\n\n1. **The TRA framework existed and was board-supervised.** The Board \"had ultimate approval\" over each TRA — defeating any defense suggestion that Microsoft set its own terms.\n2. **Microsoft's 2019 $1B → $20B (20×) TRA is a confirmed first-deal anchor.** This becomes the foundation Schizer builds on in Day 9 (see [[Day 9|Day 9 digest]]) when he characterizes the residual-after-$275B framework as \"moving the outfield fence back by 20% every year.\"\n3. **Paragraph 6 is the bombshell.** The Board \"**did not consider any financial or quantitative analysis specifically prepared by a third-party financial advisor to assess any proposed TRAs**.\" That single line is what lets Schizer say (Day 9) that **Goldman Sachs was retained for the 2025 restructuring but no comparable advisor was retained for the 2023 $10B Microsoft round** — \"**it would have been custom and practice to have a financial adviser crunching the numbers**.\"\n\nThe Court's limiting instruction (Day 8) — admitting the evidence only against OpenAI, not Microsoft — is critical: PricewaterhouseCoopers and Andersen Consulting valuation reports are now part of the OpenAI Inc. record but cannot be wielded against Microsoft directly.\n\nA note on transcription: the Day 8 wiki digest renders the second valuation firm as \"**Andreessen Consulting**.\" The official sworn declaration says \"**Andersen Consulting**\" — the older accounting consultancy (predecessor to Accenture), not Andreessen Horowitz / a16z. The Day 8 digest should be corrected on that point.\n\n---\n*See also:* [[Day 8]] · [[Day 9]] · [[Robert Wu]] · [[David Schizer]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-295", "exhibit": "PX 295", "party": "Plaintiffs", "type": "News article", "admitted_trial_day": "Day 3 (April 29, 2026)", "uploaded_box_pt": "2026-04-29T15:09:19", "uploader": "Someone", "pages": 4, "size_bytes": 4607141, "source_pdf": "PX-295.pdf", "pdf_url": "https://media.mts-in.com/PX-295.pdf", "body_markdown": "# PX 295 — Oct 20, 2022 *The Information*: \"OpenAI, Valued at Nearly $20 Billion, in Advanced Talks with Microsoft For More Funding\"\n\n> An *Information* exclusive by Aaron Holmes, Kate Clark, Erin Woo and Amir Efrati reporting OpenAI in advanced talks with Microsoft for additional funding at a ~$20B valuation, naming Musk's resignation from the board, the 100x return cap, and Altman's \"push a button and say how much money you want the company to get\" line.\n\n## Document type\n**News article (web capture, 4 pages).** Published October 20, 2022, 1:47pm PDT, on TheInformation.com. Bates 2024MUSK-0014486–0014489. Stamped \"295.001\"–\"295.004\" with the Northern District of California PLAINTIFF exhibit sticker (Case 4:24-CV-04722-YGR, Mark Busby, Clerk).\n\n## Logistics\n- **Trial admission:** Day 3 (April 29, 2026) — Musk direct concludes; cross begins.\n- **Box upload:** 2026-04-29 15:09:19 PT — Day 3 mid-afternoon batch (clustered with PX 157, PX 233, PX 355).\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~4.6 MB, 4 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `295.pdf`.\n\n## Transcribed text\n\n> **The Information**\n> *Exclusive*\n>\n> # OpenAI, Valued at Nearly $20 Billion, in Advanced Talks with Microsoft For More Funding\n>\n> *Image of Sam Altman raising money, generated by OpenAI's Dall-e 2*\n>\n> By Aaron Holmes, Kate Clark, Erin Woo and Amir Efrati\n> Oct 20, 2022, 1:47pm PDT\n> Comments by Red A, Field Garthwaite, and 4 others\n>\n> **OpenAI**, whose text- and image-generating artificial intelligence has become a mainstream hit, is in advanced talks to raise more funding from Microsoft, which previously backed the startup with capital that includes credits to use Microsoft's Azure cloud computing services to develop its technology, according to a person with knowledge of the discussions. A new deal could help Microsoft grow Azure usage, one of its top priorities, while keeping OpenAI's business away from rivals including Amazon Web Services and Google Cloud.\n>\n> The talks follow a previously undisclosed sale of OpenAI stock by existing shareholders last year to investors including **Sequoia Capital, Tiger Global Management, Bedrock Capital and Andreessen Horowitz**. In that deal, the price of the shares implied a valuation of nearly **$20 billion** for the seven-year-old startup, said several people with knowledge of the deal.\n>\n> ## The Takeaway\n> → AI startup in advanced talks for more Microsoft funding\n> → Sequoia, Tiger, Andreessen bought shares at close to $20 billion valuation\n> → The investors valued OpenAI at more than 500 times forward sales\n>\n> The quiet share purchases represented a big bet on the future of a company that former Y Combinator President Sam Altman, Tesla CEO Elon Musk and other AI practitioners founded as a nonprofit to wrest AI talent and research away from for-profit giants like Google and Facebook. It has since emerged as one of the best-known startups in a corner of AI that recently achieved breakthroughs in helping humans quickly generate original text, images and software code. OpenAI licenses its software to other AI startups, but its revenue is still modest.\n>\n> A person with direct knowledge of OpenAI's finances implied the company was on track to generate revenue in the low tens of millions of dollars this year. That means OpenAI's valuation last year likely was between 500 and 800 times the revenue it projected in 2022.\n>\n> Training machine learning models is a costly business, so OpenAI needs plenty of cash. The startup launched with **$1 billion in funding** from high-profile investors including Musk, LinkedIn co-founder Reid Hoffman, Founders Fund partner Peter Thiel and Greg Brockman, a former chief technology officer at Stripe who is now OpenAI's president and chairman. Then, in 2019, Microsoft invested in the startup in a deal worth **$1 billion**. Khosla Ventures also invested directly in the startup, according to a person with direct knowledge of the investment.\n>\n> As part of that Microsoft deal, the two companies agreed to jointly develop AI tools for Azure and other areas of Microsoft's business, and OpenAI agreed to move all of its cloud spending to Azure, where it would train its machine-learning models. That marked a blow to Google's rival cloud service, where OpenAI spent roughly **$70 million on cloud computing in each of 2019 and 2020**, according to a document viewed by The Information. (Amazon Web Services was part of a group that funded OpenAI when the startup launched in 2015 but it isn't clear how much money OpenAI spent with AWS.)\n>\n> OpenAI said it had welcomed the Microsoft investment because it gets it closer to developing what it calls an artificial general intelligence, or AGI—technology that's indiscernible from human intelligence—and that, it says, will broadly benefit humanity. By the time it had taken the Microsoft investment, the startup had turned into a for-profit company and Musk had resigned from the board. In an effort to prevent investors from driving the company to focus purely on the bottom line, OpenAI said it was capping the return on investments for its early investors to **100 times their capital**. The cap would be lower for later investors.\n>\n> \"**We fought really hard about this problem, because we were aware that if you really do make an AGI, it's basically like, push a button and say how much money you want the company to get,**\" Altman said in an interview with The Information last year.\n>\n> ## Microsoft's Edge\n>\n> Deeper ties with OpenAI would dovetail with Microsoft CEO Satya Nadella's goal of drawing more revenue from AI tools in the coming years, according to a person with direct knowledge of his plans. A key part of that strategy is infusing more AI tools into Microsoft's existing products, like 365, Teams, and its Windows operating system, the person said. Microsoft announced some of those products at its recent Ignite conference, including tools that analyze productivity in its Teams collaboration software.\n>\n> Nadella also sees AI as a key growth driver for so-called low-coding tools, which help people develop software even if they don't have sophisticated coding knowledge, the person said. Microsoft's GitHub Copilot uses OpenAI's Codex model to automatically suggest lines of code to developers as they write programs. Copilot is currently available to individual developers but the company expects to branch out to enterprise customers in the coming year.\n>\n> A number of companies have already built promising businesses from applications that run on top of OpenAI's GPT-3 technology, which can quickly help a human generate original text on a topic of their choosing. The Information earlier this week reported that one of those companies, **Jasper AI**, which helps marketers produce text for blog posts or advertisements, has targeted annualized revenue of **$80 million** by the end of this year, up from around $30 million in annualized revenue a year earlier—the first year it started generating revenue. Annualized revenue, also known as a run rate, is a measure of the prior month's revenue multiplied by 12. Jasper AI says its customers include HarperCollins and Airbnb.\n>\n> Amid its torrid growth, Jasper garnered a private valuation of more than **$1.5 billion** from investors this summer. It isn't clear what percentage of that revenue Jasper pays back to OpenAI, but OpenAI earlier this year reduced its fees for companies using GPT-3 by two-thirds, and investors believe OpenAI is purposely charging a lower rate in order to increase adoption, according to a person with knowledge of the company.\n>\n> One factor that could gum up OpenAI's ambitions is the rise of open source AI software. OpenAI captured the internet's imagination earlier this year when it released **Dall-e 2**, which produces original images when someone types a simple text description of what they want to see, like \"teddy bears working on new AI research on the moon in the 1980s.\" But another text-to-image creation platform called **Stable Diffusion** has since burst onto the scene, much to the surprise of the AI world. Using Stable Diffusion, startups including Stability AI and Runway ML have developed content creation tools for their customers.\n>\n> Google and Facebook owner Meta Platforms also have shown similar text-, image- and video-generating AI made by their research groups, but it isn't clear if or when those companies will try to commercialize them or offer them for public use. Altman earlier this year said OpenAI is readying GPT-4, an update to the GPT-3 text-generation tool that the company released in 2020.\n>\n> ---\n> *Jon Victor contributed to this article.*\n>\n> *Aaron Holmes is a reporter covering tech with a focus on enterprise and cybersecurity. … Kate Clark is a deputy bureau chief at The Information and the author of the twice-weekly column, Dealmaker. … Erin Woo is a San Francisco-based reporter covering Google and Alphabet for The Information. … Amir Efrati is executive editor at The Information, which he helped to launch in 2013. …*\n>\n> Conversation — 6 comments\n> Charlie Christoffersen — Post a comment\n\n## Commentary\n\nPX 295 is a contemporaneous public-record artifact plaintiffs use to fix the world's awareness in **October 2022** of OpenAI's Microsoft-anchored capital structure, the **100x return cap**, and Altman's own famous \"push a button and say how much money you want the company to get\" framing — i.e., a pre-ChatGPT-launch news story that already names the structural facts plaintiffs say betrayed the charity. For the [[Key Themes#\"Captured by Microsoft\" (statute-of-limitations fight)|statute-of-limitations fight]], the article is double-edged: it predates Musk's Sept. 2020 \"captured by Microsoft\" tweet by two years on the underlying Microsoft-Azure exclusivity facts, but it also predates ChatGPT's Nov. 2022 launch and the November 2023 Altman firing — the events Musk has tied to \"phase three\" disillusionment ([[Key Themes#Musk's \"three phases\"|three phases]]). The Altman quote is the cleanest contemporaneous public statement of the [[Quotes|\"$1B\"]] / 100x-cap mindset that birthed the 2019 capped-profit OpenAI LP and that Birchall later called \"**posturing and virtue signalling**\" on the [[Key Themes#The 2018 \"purple box\" term sheet|purple-box]] direct. Used on Day 3 during the close of Musk's direct, the article belongs to plaintiffs' set piece showing what was publicly visible — and what was not — about OpenAI's drift toward a for-profit before the 2023 lawsuit.\n\n---\n*See also:* [[Day 3|Day 3 digest]] · [[Sam Altman]] · [[Key Themes]] · [[Quotes]]\n"}
{"exhibit_id": "PX-296", "exhibit": "PX 296", "party": "Plaintiffs", "type": "Text messages", "admitted_trial_day": "Day 3 (April 29, 2026)", "uploaded_box_pt": "2026-04-29T15:09:18", "uploader": "Someone", "pages": 3, "size_bytes": 170830, "source_pdf": "PX-296.pdf", "pdf_url": "https://media.mts-in.com/PX-296.pdf", "body_markdown": "# PX 296 — Oct 23, 2022 Musk-Altman \"bait and switch\" text thread\n\n> The two-party text thread on the morning of Musk's Twitter close-out trip — Musk's \"**This is a bait and switch**\" complaint about OpenAI's $20B valuation, Altman's apologetic equity-offer reply, and Musk's commitment to meet \"Tues or Wed.\"\n\n## Document type\n**Text messages, Cellebrite-style \"Short Message Report\" extraction.** Three pages: cover sheet (1 conversation, 10 messages, 5 participants — Elon Musk and Sam Altman, with handles partially redacted) plus two pages of message-by-message transcript with timestamps in GMT-07:00. Bates 2024MUSK-0006375–6377.\n\n## Logistics\n- **Trial admission:** Day 3 (April 29, 2026). Per [[Key Themes]]: \"PX 296 — Oct 2022 bait-and-switch text.\" Used during Musk's direct/cross to anchor the timing of his \"phase three\" disillusionment narrative — Musk dates his \"extreme discomfort\" to \"late '22 and '23\" (042926TT.txt:7148).\n- **Box upload:** 2026-04-29 15:09:18 PT — Day 3 mid-afternoon batch (clustered with PX 157, PX 233, PX 355).\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~167 KB, 3 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `296.pdf`.\n\n## Transcribed text\n\n> **Conversation:** [chat — handles partially redacted]\n> **Date Range:** 10/23/2022\n> **Participants:** Elon Musk, Sam Altman\n>\n> **Messages in chronological order (GMT -07:00):**\n>\n> **EM** (Elon Musk) — 10/23/2022, 6:06 AM\n> Elon here\n>\n> **EM** — 10/23/2022, 6:07 AM\n> New Austin number\n>\n> **EM** — 10/23/2022, 6:08 AM\n> I was disturbed to see OpenAI with a $20B valuation. De facto. I provided almost all the seed, A and most of B round funding.\n>\n> **EM** — 10/23/2022, 6:08 AM\n> https://www.theinformation.com/articles/openai-valued-at-nearly-20-billion-in-advanced-talks-with-microsoft-for-more-funding\n> [attachments: 4 KB and 554 KB pluginPayload attachments]\n>\n> **EM** — 10/23/2022, 6:08 AM\n> [attachment ID]\n>\n> **EM** — 10/23/2022, 6:08 AM\n> [attachment ID]\n>\n> **EM** — 10/23/2022, 6:08 AM\n> **This is a bait and switch**\n>\n> **SA** (Sam Altman) — 10/23/2022, 10:47 PM\n> I agree this feels bad—we offered you equity when we established the cap profit, which you didn't want at the time but we are still very happy to do any time you'd like.\n>\n> We saw no alternative, given the amount of capital we needed and needing still to preserve away to 'give the AGI to humanity', other than the capped profit structure.\n>\n> Fwiw I personally have no equity and never have. Am trying to navigate tricky tightrope the best I can and would love to talk about how it can be better any time you are free. Would also love to show you recent updates.\n>\n> **EM** — 10/23/2022, 10:51 PM\n> I will be in SF most of this week for the Twitter acquisition. Let's talk Tues or Wed.\n>\n> **SA** — 10/23/2022, 10:51 PM\n> Liked \"I will be in SF most of this week for the Twitter acquisition. Let's talk Tues or Wed.\"\n\n## Commentary\n\nPX 296 anchors a moment plaintiffs needed precisely on the calendar: Musk's documented use of the **\"bait and switch\"** label *before* he filed suit, and dated to October 23, 2022 — i.e., before ChatGPT (released November 30, 2022) and before the November 2023 Altman firing. That timestamp matters because it puts a contemporaneous Musk complaint into the SOL period. Two other facts on the page are worth noting. Altman's reply (\"**we offered you equity when we established the cap profit, which you didn't want at the time but we are still very happy to do any time you'd like**\") is a direct admission that Musk had been offered equity in the LP and refused, and Altman's \"**Fwiw I personally have no equity and never have**\" tracks his contemporaneous public stance and the August 2018 LP term sheet cover email ([[DX 827]]: \"my current thought is that I won't take any equity\"). The texting context — Musk was in San Francisco for the Twitter deal — also explains why Musk closed the thread with a meeting offer rather than escalation. Notable that this exchange occurred on the same Sunday Musk was juggling the Twitter close-out; the plaintiffs' theory of progressive disillusionment depends on small contemporaneous receipts like this one.\n\n---\n*See also:* [[Day 3|Day 3 digest]] · [[Sam Altman]] · [[Phases One Two Three]] · [[DX 827]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-301", "exhibit": "PX 301", "party": "Plaintiffs", "type": "Slack messages", "admitted_trial_day": "Day 8 (May 6, 2026)", "uploaded_box_pt": "2026-05-06T14:07:58", "uploader": "Someone", "pages": 2, "size_bytes": 819508, "source_pdf": "PX-301.pdf", "pdf_url": "https://media.mts-in.com/PX-301.pdf", "body_markdown": "# PX 301 — Murati ↔ Kwon Slack DM re: GPT-4 Turbo \"DSB\" review (Sam negotiating directly with Satya)\n\n> A two-page Slack DM screenshot between Mira Murati and Jason Kwon flagging that Sam Altman was negotiating directly with Satya Nadella while bypassing Deployment Safety Board review for GPT-4 Turbo.\n\n## Document type\n**Text messages (Slack DM screenshots).** Two pages, each a phone screenshot of a Slack one-on-one chat (\"Jason Kwon\") at 7:29 device time. Messages timestamped between 9:44 PM and 10:41 PM. Bates SUTSKEVER_MUSKSUB_00000532–533. Marked HIGHLY CONFIDENTIAL — ATTORNEYS' EYES ONLY (declassified to the public Box upon admission).\n\n## Logistics\n- **Trial admission:** Day 8 (May 6, 2026) — used during the Murati live direct examination. No prior wiki reference; admission day inferred from Box upload date.\n- **Box upload:** 2026-05-06 14:07:58 PT — Day 8 early-afternoon batch.\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~800 KB, 2 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `301.pdf`.\n\n## Transcribed text\n\n> **Slack DM with Jason Kwon**\n>\n> **Mira Murati** — 9:44 PM\n> Hi Jason, couple of questions for you:\n> - on dsb, is miles aware that Sam is negotiating this directly with Satya?\n>\n> *(3 replies in thread)*\n>\n> **Mira Murati** — 9:45 PM\n> - On turbo going through dsb, I don't really want to argue with people and it takes no real extra effort to put together the review docs for it but Sam told me that you had said it doesn't need to per legal. Is this correct?\n>\n> **Jason Kwon** — 10:37 PM\n> ugh\n>\n> **Jason Kwon** — 10:37 PM\n> I need to talk to Sam - I wasn't aware that he was negotiating this right this minute with Satya, just that he had a general principle he wanted to follow.\n>\n> *(1 reply)*\n>\n> **Jason Kwon** — 10:39 PM\n> On Turbo - I actually said something different, which was (1) that people were saying it needed DSB but that also it didn't seem like that much work and that it would be handled and (2) we should going forward not let anyone just decide something goes through DSB, but instead have a lawyer (doesn't have to be me, probably prefer it that way, maybe Che) be the one to make the determination, since it's a question of interpretation of the contract.\n>\n> *(1 reply)*\n>\n> **Jason Kwon** — 10:41 PM\n> Looking at what happened on the turbo thread, it seems like an unhealthy dynamic that the people with the power get to decide when it gets used. Just commenting on the dynamic, not the specific people. I would've had a stronger POV on whether it was necessary had someone said this is going to be hard/lot of work, instead of something that's a light lift.\n>\n> *(an inline screenshot of an earlier \"Jason Kwon\" message in the same thread is visible but text not legible in the Slack export)*\n\n## Commentary\n\nPX 301 is a Day-8 Murati-direct exhibit going to internal Deployment Safety Board (DSB) governance — the very mechanism Altman described in [[PX 25]] (Nov. 2015): \"any technology that potentially compromises the safety of humanity has to get consent of the Board to be released.\" Plaintiffs use this thread to show Altman bypassing those internal safety controls — \"**Sam is negotiating this directly with Satya**\" while purportedly authorizing GPT-4 Turbo to skip DSB review. Kwon's \"**unhealthy dynamic that the people with the power get to decide when it gets used**\" is the live-witness sound-bite. Cross-reference: this exhibit pairs with the Toner video deposition opening on Day 8 (in which Toner — a former OpenAI board member — addresses the November 2023 board crisis triggered partly by safety-process concerns).\n\n---\n*See also:* [[Day 8|Day 8 digest]] · [[Mira Murati]] · [[PX 25]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-302", "exhibit": "PX 302", "party": "Plaintiffs", "type": "Memo", "admitted_trial_day": "Day 8 (May 6, 2026)", "uploaded_box_pt": "2026-05-06T14:07:58", "uploader": "Someone", "pages": 2, "size_bytes": 2169844, "source_pdf": "PX-302.pdf", "pdf_url": "https://media.mts-in.com/PX-302.pdf", "body_markdown": "# PX 302 — Sept 30, 2022 Mira Murati feedback memo to Sam Altman (\"only Sam had access to this\")\n\n> Mira Murati's six-section structured memo to Altman flagging a deteriorating internal management culture: \"alignment with Sam & exec on company strategy & goals,\" \"alignment on Applied's role,\" \"focus on our users,\" \"safety policies & risk management,\" \"moving slowly,\" and \"things Sam can do to help\" — including \"set up in-depth reviews until you are satisfied that you understand the situation,\" \"Sam saying one thing to one person and completely opposite to another,\" and a flag that Altman had been blaming employees for the DALL-E \"what went wrong.\"\n\n## Document type\n**Memo — internal performance feedback document.** Two pages, photographed off a screen (likely Murati's laptop). Header: \"Sept 30, 2022 Feedback from Mira to Sam (only Sam had access to this).\" Bates SUTSKEVER_MUSKSUB_00000565–566.\n\n## Logistics\n- **Trial admission:** Day 8 (May 6, 2026) — used during the live direct examination of Mira Murati. The exhibit corresponds to the \"screenshot is not about the safety of GPT-4 Turbo. It's about Sam\" line at *(5/6/2026 Testimony @ ~10:02 PT)* in [[Key Themes]] §\"Day 8 — Sam Altman's candor.\"\n- **Box upload:** 2026-05-06 14:07:58 PT — Day-8 morning batch.\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~2.1 MB, 2 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `302.pdf`.\n\n## Transcribed text\n\n> **Sept 30, 2022 Feedback from Mira to Sam (only Sam had access to this)**\n>\n> **Problems**\n>\n> **Alignment with Sam & Exec on company strategy & goals**\n>\n> We have goal setting, quarterly planning and project planning in place for the company. The constant panic around our projects, people, goals etc generates chaos and churn. We talk about focus but in practice our approach is do-everything and do it fast because we constantly get pressure to change priorities and shuffle around people and projects.\n>\n> *Proposed solution:*\n> * We agree on projects and goals on a regular cadence. When things come up, please go through me or my direct reports, not people that don't have the whole picture (basically the people that at the end of the day you would fire if things didn't go well).\n> * As a result of trying to move very fast, we are not very good at documentation. I make tens of decisions every single day small and big but they're not documented so when we look back, the tradeoffs are in my head or with other stakeholders. Because you're not very involved operationally, you're not aware of all the tradeoffs and sometimes you display disappointment about what didn't get done even if there was a great reason for it. For example, a tradeoff we made was to launch dall-e vs reforming T&S policies, we didn't have staffing to do both. It would slow down execution to try documenting everything but I can pick top 3 decisions each week and document them to ensure we're in sync about tradeoffs.\n> * Between us and exec we align on priorities and things we're not going to do in written form and try to stick to them for 3 months at a time, unless there's a good reason to reconsider.\n> * Overall bringing more transparency on execution of the company to you and Greg seems important to avoid situations where you make assumptions about what the issues are. Mira will prepare weekly updates for you that span the company's progress on everything, so that we can have reasonable and productive conversations about how to move faster and more effectively.\n>\n> **Alignment with Sam & Exec on Applied's role**\n>\n> Applied work at OpenAI from the beginning has happened through sheer power of will — I strongly believe it's a necessary strategic piece to get to AGI and figure out how to make it go well for the world, I don't see it as an intermediary tool to get the funding we needed from MSFT. I feel that you don't have the same conviction because every couple of months, we talk about transitioning Applied to **msft**, splitting it off, the fund taking over the function or getting rid of it all together. This is important because without conviction, we can only get as far as the next funding goal and cannot make it through the hard times. Whenever we face a difficult situation, the whole strategy gets called into question because we're not aligned on the long term importance of Applied.\n>\n> *Proposed solution:* Use your staff meeting or another time to have an informed discussion with key stakeholders and settle this question. I propose that you ask people with strong opinions to write a document. As a result have the final outcome integrated in the Applied charter.\n>\n> **Focus on our users**\n>\n> Tactically I think we can resource Applied much better and with appropriate talent to be even more customer focused but the fundamental issues here are (1) cultural and (2) strategic.\n>\n> 1. Doing what the users want is not in the DNA of OpenAI. OpenAI started as a company that would protect humanity from the use of AI. Since we've created the product team, we've met a lot of internal and external resistance in building products and deploying them, one of these tensions (Anthropic split) almost put the company in an existential crisis. In the aftermath of Anthropic we've worked hard to build trust both internally and externally with the public. The strategies we've pursued on iterative deployment have been critical to building public trust, encouraging people to join us versus Anthropic or Brain or DM. We have been on a defense mode on why we're building products and why we deploy them. Deployment is seen as bad and therefore users are seen as bad. This is not an Applied problem, it's a whole company cultural problem. We have to defend to the board our decisions on deployment.\n> 2. This year the most cited goal to the company has been the **$100M revenue goal**. You said that Kevin and Satya said this was needed to raise the next $10B — it didn't matter how we got to this number, we needed to get there. We pressured the GTM team to do big quick contracts instead of carrying out a cohesive strategy around the platform.\n>\n> *Proposed solution:*\n> 1. I think that as a founder CEO you're uniquely suited to shape this change. Saying it explicitly at all hands that this is a major cultural shift and why we're doing it is important. In practice then supporting decisions that are in line with this will be even more important.\n> 2. When you think it's necessary to change goals for the whole company, let's discuss them in your staff meeting or another meeting with the right stakeholders to ensure we understand the risks and align on them. We need to play out the consequences, asking people what would worry them about the particular decision or how it could play out in a way that we don't want to make sure we've really considered the downsides. With the open source discussion, we managed to do this fairly well although it was a bit scattered and disoriented. This might delay our execution by a couple of days but in the long run it will save us time. I am happy to help set agendas and prepare the information to have a productive meeting where we reach a decision.\n>\n> **Safety policies & risk management**\n>\n> Our culture on safety was set in the early founding days, cemented with GPT-2 release later, and incremental pushes are not sufficient to make the changes we need on reasonable safety and risk management. We need a complete overhaul here where the company understands the approach and most importantly trusts the leaders to make the call.\n>\n> *Proposed solution:* You tell the company concretely how we will handle safety & risk moving ahead, Mira in charge of all calls (unless Jason thinks my decision is sending us to jail) + Mira will work out with Peter how to **operationalize** it. It's important for you to stay engaged here at least initially and that I have your support so I am not constantly fighting with Steve or Greg or whoever may have a completely different view even if they don't see the global picture like I do.\n>\n> **Moving slowly**\n>\n> We have grown roughly 2x through covid and we've moved very fast, tried to do as many things as possible. A lot of things are broken, there's too much friction. We don't have central knowledge management, documentation, a lot of things at OpenAI run through know-how. Infrastructure is fragmented, we have made a lot of decisions that locally have helped people and teams move fast but now we're paying the price. **Torchflow** usability is a great example of this.\n>\n> *Proposed solution:* Don't panic. Things like superassistant team design through my time out of the window. Let's take a hit over the next 2 months to fix the foundations across our infra, team design and people.\n>\n> **Things Sam can do to help:**\n>\n> 1. Actively get informed and use our official meetings to do that effectively. If the current channels don't work, let's make new ones.\n>    1. Often I hear from you two things simultaneously, that to me seem in conflict: (1) We're not moving fast enough or a particular area or person is failing & (2) You don't know what's going on, so you might be wrong. When unsure of how things are going or if there's a feeling that things are not going well, go directly to Mira to get information and **set up in-depth reviews until you are satisfied that you understand the situation.** You rarely ask me what I think are the big problems and where we need support and that's fine when you're not operationally involved. In the past few months, you have gotten more operationally involved and in this scenario it is very important that you beliefs are not formed superficially but you take the time to really understand what the issues are.\n>    2. In the past couple of months, you and Greg have been telling the company in informal channels and all-hands that we have failed with DALL-E, without doing the work to understand what went wrong here, what failing means, and what needs to happen differently. **This erodes trust in leadership, including you because it feels that when things go wrong we blame the employees instead of putting in the work to understand what has happened and what needs to change.** Critiques can be powerful and motivating but they need to be grounded and offer a path forward especially when they come from the CEO. To figure out the path forward, we need to have an informed discussion with people doing the work on the ground. It's not productive to send tactical direct guidance to Aditya or Joanne, it confuses them and creates weird dynamics where Joanne needs to manage up to the CEO of the company when she's just a few years into her career and has no ability to do that whatsoever.\n> 2. The official channels of discussion have failed. These important conversations don't come up in your staff meeting so you end up having pairwise conversations with people.\n>    1. For example, Applied is my area of responsibility at the end of the day but before you talked with me in depth about the issues, you already had your mind made up on what to do (kill it or move it under Brad) from conversations with Greg and Brad and I had to reverse engineer where this was coming from. This is my life's work, I take your ideas, suggestions, criticisms very seriously and I feel sabotaged and completely unsupported when this happens. It feels that I am working against you to make something successful, in spite of your direction, versus relying on you as a trusted advisor and finding support to fix difficult problems across the **organization**.\n>    2. If Greg is an important advisor to you, let's add him to your staff meeting where we can all discuss topics in the light of day versus side conversations that create politics, mistrust and misunderstandings.\n> 3. Talk about your concerns with me directly. I don't want to find out from others that you thought that the DV3 plan should have taken 2 days instead of 2 weeks. It's a missed opportunity for us to resolve important issues for the company and it undermines the leadership of the company when you do this.\n\n## Commentary\n\nPX 302 is the documentary backbone of [[Mira Murati]]'s Day 8 testimony — and one of the trial's clearest contemporaneous records of internal management dysfunction at OpenAI in 2022. Plaintiffs use it for two purposes: (i) **to corroborate Murati's \"Sam was not always candid\" testimony** with her own pre-departure writing, neutralizing defense's \"axe-grinding\" frame (see [[Key Themes]] §\"Day 8 — Sam Altman's candor\"); and (ii) to anchor the broader theme that Altman's stewardship of OpenAI Inc.'s charitable mission was failing in real time, more than a year before the November 2023 firing. The \"this year the most cited goal to the company has been the $100M revenue goal… needed to raise the next $10B\" passage is plaintiffs' best evidence that OpenAI's day-to-day priorities had quietly transitioned from \"primary fiduciary obligation… to advance the aims of the OpenAI Charter\" ([[PX-239]]) to a Microsoft-driven revenue ramp. Note also Murati's frank statement: \"Doing what the users want is not in the DNA of OpenAI. OpenAI started as a company that would protect humanity from the use of AI.\" That line — written by the then-CTO — is the cleanest internal articulation of the original mission anywhere on the trial record.\n\n---\n*See also:* [[Day 8|Day 8 digest]] · [[Mira Murati]] · [[Sam Altman]] · [[Greg Brockman]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-304", "exhibit": "PX 304", "party": "Plaintiffs", "type": "Court filing", "admitted_trial_day": "Day 8 (May 6, 2026)", "uploaded_box_pt": "2026-05-05T16:29:57", "uploader": "Someone", "pages": 5, "size_bytes": 596437, "source_pdf": "PX-304.pdf", "pdf_url": "https://media.mts-in.com/PX-304.pdf", "body_markdown": "# PX 304 — Nov 16, 2023 OpenAI Inc. Board Unanimous Written Consent firing Sam Altman\n\n> The five-page DocuSigned Unanimous Written Consent of the OpenAI Inc. Board of Directors, dated November 16, 2023, that **terminated Sam Altman's employment**, removed him as President and CEO of OpenAI Inc., OpenAI GP, and the Managed Entities, removed his signatory authority, and **appointed Mira Murati as Interim CEO** — the formal corporate document underlying the November 2023 firing that runs through Day 8 of the trial.\n\n## Document type\n**Corporate written consent (board resolution), DocuSigned, five pages.** Adopted under Article IV, Section 11 of the OpenAI Inc. Bylaws (eff. Jan. 4, 2016) and Section 141(f) of the Delaware General Corporation Laws. Signed by all four non-employee directors: **Adam D'Angelo, Tasha McCauley, Ilya Sutskever, Helen Toner**, all 11/16/2023. DocuSign Envelope ID: F4668B42-181D-4725-B5A3-6136166FE727. Bates OPENAI_MUSK00027400–00027404. Stamped \"Highly Confidential.\"\n\n## Logistics\n- **Trial admission:** Day 8 (May 6, 2026) — the day Murati and Zilis testified live and Toner's video deposition began. Used as the documentary spine for plaintiffs' November-2023-firing thread (see [[Key Themes]] § \"Microsoft's role in Altman's reinstatement\" and § \"Day 8 — Sam Altman's candor\").\n- **Box upload:** 2026-05-05 16:29:57 PT — late-afternoon Day 7 batch, ~24 hours before Day 8 admission.\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~582 KB, 5 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `0204.pdf`. (Note: the Box filename `0204.pdf` predates assignment of the PX 304 trial-exhibit number.)\n\n## Transcribed text\n\n> **UNANIMOUS WRITTEN CONSENT OF THE BOARD OF DIRECTORS OF OPENAI, INC.**\n>\n> The undersigned, being all of the members of the Board of Directors (the \"Board\") of OpenAI, Inc., a Delaware nonprofit nonstock corporation (the \"Corporation\"), and acting in accordance with Article IV, Section 11 of the Bylaws of the Corporation, effective as of January 4, 2016 (the \"Bylaws\"), and Section 141(f) of the Delaware General Corporation Laws, do hereby adopt, and do consent to the adoption of, the following resolutions in its capacity as the Board of the Corporation:\n>\n> ## Removal of Sam Altman as Chief Executive Officer\n>\n> WHEREAS, pursuant to Article VI, Section 3 of the Bylaws, the Board has the authority to remove any officer of the Corporation with or without cause;\n>\n> WHEREAS, the Corporation is the sole member and the managing member of OpenAI GP, L.L.C. (formerly known as SummerSafe GP, L.L.C.), a Delaware limited liability company (\"OpenAI GP\");\n>\n> WHEREAS, (1) OpenAI GP serves as the manager for OpenAI Global, LLC, a Delaware limited liability company (\"OpenAI Global\") and for a number of entities affiliated with OpenAI Global, including, without limitation, Aestas Management Company, LLC, a Delaware limited liability company, Aestas, LLC, a Delaware limited liability company, and OpenAI Holdings, LLC, a Delaware limited liability company; (2) OpenAI Holdings LLC, a Delaware limited liability company, is the sole stockholder of OAI Corporation, a Delaware corporation, and (3) OpenAI Global is the sole member and manager of OpenAI OpCo, LLC (formerly known as OpenAI, L.P.), a Delaware limited liability company (collectively, the \"Managed Entities\");\n>\n> WHEREAS, Sam Altman is currently employed by and serving as the President and Chief Executive Officer of the Corporation, and as Chief Executive Officer of OpenAI GP and certain of the Managed Entities; and\n>\n> WHEREAS, **the Board has lost trust in Sam Altman's ability to be candid and forthright in his communications with the Board and employees of the Corporation and OpenAI Global, LLC, and is concerned about the resulting impact of his actions on the Corporation's mission**, and, as a result, the Board desires to terminate Mr. Altman's employment with the Corporation and OpenAI GP, to remove him as the President and Chief Executive Officer of the Corporation, as Chief Executive Officer of OpenAI GP and the Managed Entities, as well as from any other office in which he may serve for the Corporation, OpenAI GP and the Managed Entities, and to remove any signatory authority on behalf of the Corporation, OpenAI GP or the Managed Entities previously granted to him.\n>\n> NOW, THEREFORE, BE IT RESOLVED, that the Board hereby, effective immediately, terminates Mr. Altman's employment with the Corporation and OpenAI GP, removes Mr. Altman from any position in which he serves for the Corporation or for OpenAI GP, including as President and Chief Executive Officer of the Corporation and as Chief Executive officer of OpenAI GP, removes any signatory authority on behalf of the Corporation or OpenAI GP previously granted to him, and directs OpenAI GP as the manager of certain of the Managed Entities to take all actions necessary to terminate his employment, effective immediately, remove him from any office in which he serves with any of the Managed Entities and remove any signatory authority on behalf of any of the Managed Entities previously granted to him.\n>\n> ## Appointment of Mira Murati as Interim Chief Executive Officer\n>\n> WHEREAS, the Board desires to appoint Mira Murati as the Interim Chief Executive Officer of the Corporation, OpenAI GP and the Managed Entities.\n>\n> NOW, THEREFORE, BE IT RESOLVED, that the Board hereby appoints Ms. Murati as the Interim Chief Executive Officer of the Corporation and of OpenAI GP, and directs OpenAI GP as the manager of certain of the Managed Entities to take all actions necessary to appoint Ms. Murati as the Interim Chief Executive Officer of the Managed Entities, effective immediately, and to hold such office until her successor has been appointed and qualified, or her earlier death, resignation, removal or disqualification.\n>\n> ## General Authorizing Resolutions\n>\n> RESOLVED, FURTHER, that the directors of the Corporation (each, an \"Authorized Representative\") be, and each of them hereby is, authorized, empowered and directed, for and on behalf of the Corporation, to take any and all actions, to negotiate for and enter into agreements and amendments to agreements, to perform all such acts and things, to execute, file, deliver or record in the name and on behalf of the Corporation, all such certificates, consents, approvals, instruments, agreements or other documents, and to make all such payments as they, in their judgment, or in the judgment of any one or more of them, may deem necessary, advisable or appropriate in order to carry out the purpose and intent of the foregoing resolutions [...]\n>\n> RESOLVED FURTHER, that this written consent shall become effective immediately following the effectiveness of the **Written Consent of a Majority of the Members of OpenAI, Inc. adopted November 16, 2023 attached hereto as Exhibit A.**\n>\n> *[Signature Page Follows]*\n>\n> **IN WITNESS WHEREOF**, each of the undersigned has executed this Consent on the date set forth opposite such person's name in such person's capacity as a director of the Corporation.\n>\n> Dated: 11/16/2023 — **Adam D'Angelo, Director** [DocuSigned]\n> Dated: 11/16/2023 — **Tasha McCauley, Director** [DocuSigned]\n> Dated: 11/16/2023 — **Ilya Sutskever, Director** [DocuSigned]\n> Dated: 11/16/2023 — **Helen Toner, Director** [DocuSigned]\n>\n> *[Signature Page to Unanimous Written Consent of the Board of OpenAI, Inc.]*\n>\n> **EXHIBIT A — WRITTEN CONSENT OF A MAJORITY OF THE MEMBERS OF OPENAI, INC.** [see attached]\n\n## Commentary\n\nPX 304 is the formal corporate instrument that triggered every subsequent fact pattern Toner, Murati, and Zilis describe on Day 8. Two passages do most of the work. First, the **\"lost trust in Sam Altman's ability to be candid and forthright in his communications with the Board\"** finding — drafted by Wachtell or D'Angelo's counsel and signed by all four independent directors — is the exact language Toner returns to in her video deposition on Day 8 (~14:36–15:05 PT) when she explains the board's reasoning. Murati's \"Was [Mr. Altman] candid with you?\" – \"**Not always.**\" testimony (~09:32 PT) tracks the same finding from inside management. Plaintiffs use this consent as the **board-level fact-finding** that makes Brockman's Charter-2.0/\"primary fiduciary duty is to humanity\" line ([[Key Themes]] § \"Charter 2.0\") look like a structural cover-up of an Altman candor problem the nonprofit's own board had already adjudicated. The four-director sign-off — D'Angelo, McCauley, Sutskever, Toner — is also the documentary basis for the Day 8 \"Microsoft's announcement created a threat\" line of testimony: this consent was unanimous, it was effective immediately, and it was reversed within days under Microsoft pressure (per Murati and Toner). Defense will argue the consent is exactly the **mission-protection lever** working as designed and that its reversal is irrelevant to the 2017–2018 charitable-trust claims at issue. Plaintiffs argue the reversal is the \"captured by Microsoft\" tweet ([[PX-251]]) coming true in real time.\n\n---\n*See also:* [[Day 8|Day 8 digest]] · [[Mira Murati]] · [[Helen Toner]] · [[Shivon Zilis]] · [[Sam Altman]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-306", "exhibit": "PX 306", "party": "Plaintiffs", "type": "Text messages", "admitted_trial_day": "Day 8 (May 6, 2026)", "uploaded_box_pt": "2026-05-06T14:07:57", "uploader": "Someone", "pages": 5, "size_bytes": 651277, "source_pdf": "PX-306.pdf", "pdf_url": "https://media.mts-in.com/PX-306.pdf", "body_markdown": "# PX 306 — Mira Murati ↔ Satya Nadella iMessage thread, November 17–21, 2023 (the Altman-firing weekend)\n\n> Five pages of iMessages between OpenAI's then-CTO and Microsoft's CEO during the four days that began with the Friday-afternoon firing of Sam Altman and ended with his return — including Murati's \"**I'm not putting my name on this**\" and her ask of Nadella to make a public statement so they \"**don't lose researchers to Demis or Elon**.\"\n\n## Document type\n**Text messages** — iMessage screenshots between Mira Murati and Satya Nadella, exported as a 5-page PDF with timestamps. Bates MSFT_MUSK000065692 onward. Marked HIGHLY CONFIDENTIAL.\n\n## Logistics\n- **Trial admission:** Day 8 (May 6, 2026) — Murati live testimony. This is the documentary record behind the [[Key Themes]] §\"Microsoft's role in Altman's reinstatement\" finding that \"Microsoft and Satya in particular were important in unifying the team\" (Murati, 5/6 @ ~10:13 PT).\n- **Box upload:** 2026-05-06 14:07:57 PT — Day 8 mid-afternoon Plaintiffs' batch (one second after [[PX 244]]).\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~636 KB, 5 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `306.pdf`.\n\n## Transcribed text\n\n> **Friday, November 17, 2023**\n>\n> **3:33:37 PM — Satya Nadella:** Hi Mira… call when you have a chance. Thx Satya\n>\n> **Saturday, November 18, 2023**\n>\n> **9:06:38 AM — Mira Murati:** Satya could you please call me when you have a chance so we can coordinate how to talk with Sam here\n>\n> **Sunday, November 19, 2023**\n>\n> **7:40:45 PM — Mira Murati:** Can you join meeting with board please\n>\n> **7:52:51 PM — Mira Murati:** I'm not putting my name on this\n>\n> **7:55:46 PM — Mira Murati:** Quick call?\n>\n> **Monday, November 20, 2023**\n>\n> **1:16:39 AM — Mira Murati:** Hi Satya, I know it's super late. Need to call you urgently\n>\n> **1:20:21 AM — Mira Murati:** The petition we're sending says the following per your conversation with Sam and we wanted to confirm. \"**Microsoft has assured us that there are positions for all OpenAI employees with the same compensation at this new subsidiary should we choose to join.**\"\n>\n> **1:20:52 AM — Mira Murati:** We're removing it from public doc and communicating it privately\n>\n> **7:10:24 AM — Satya Nadella:** Thx. So sorry. But that is right.\n>\n> **10:44:43 AM — Mira Murati:** Satya could you please make a public statement soon that shows support for the joint openai team, basically bringing the team together? It's very important that we don't lose researchers to **Demis or Elon**. The technical team is being dragged in so many recruiting directions and a unified front would help immensely\n>\n> **Tuesday, November 21, 2023**\n>\n> [Pages 2–5 of the PDF continue the thread through Nov 21, 2023, when Altman returns to OpenAI. The substantive content is the coordination between Murati and Nadella around (a) confirming the language of the staff petition that \"Microsoft has assured us that there are positions for all OpenAI employees with the same compensation at this new subsidiary should we choose to join,\" (b) Murati's request for a public Microsoft statement of support for \"the joint openai team\" so the technical staff would not \"lose researchers to Demis or Elon,\" and (c) coordination on the timing of the public reset.]\n>\n> [Verbatim transcription of pages 2–5 omitted from this entry — refer to the PDF for the day-by-day timestamped messages used to corroborate the Day 8 testimony.]\n\n## Commentary\n\nPX 306 is Plaintiffs' single best document for the [[Key Themes]] §\"Microsoft's role in Altman's reinstatement\" theory: Microsoft (Nadella personally) confirmed in writing the offer to hire away the entire OpenAI technical team — the very leverage Toner described on Day 8 as \"**a threat**\" to the board's ability to carry out its mission. Murati's \"**I'm not putting my name on this**\" line is also the direct contemporaneous version of what she repeated on the stand about Altman's management — see the Murati quotes in [[Key Themes]] §\"Murati on Altman's management\" (e.g., \"Was [Mr. Altman] candid with you? — Not always.\"). The \"we don't lose researchers to **Demis or Elon**\" reference is the live-fire moment in which Musk's name surfaces as a recruiting alternative during the firing weekend, six weeks before Musk filed this lawsuit. Pair with [[PX 244]] for the 2019 Microsoft-OpenAI commercial-IP backdrop and with the Toner/McCauley video depositions for the directors' side of the same firing.\n\n---\n*See also:* [[Day 8|Day 8 digest]] · [[PX 244]] · [[Mira Murati]] · [[Helen Toner]] · [[Sam Altman]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-309", "exhibit": "PX 309", "party": "Plaintiffs", "type": "Text messages", "admitted_trial_day": "Day 8 (May 6, 2026)", "uploaded_box_pt": "2026-05-06T14:07:57", "uploader": "Someone", "pages": 2, "size_bytes": 831482, "source_pdf": "PX-309.pdf", "pdf_url": "https://media.mts-in.com/PX-309.pdf", "body_markdown": "# PX 309 — Nov 17–21, 2023 Murati ↔ [Microsoft contact] firing-week texts\n\n> A two-page iMessage transcript of a Mira Murati (\"openai\") ↔ a Microsoft-side contact text exchange spanning **November 17–21, 2023** — i.e., from the Friday afternoon of Sam Altman's firing through the Tuesday-night reinstatement announcement — including Murati's \"**Don't worry Kevin, we resolved. Close to having the board resign**\" and final \"**We've reached agreement. Sam will return as CEO with a new initial board Bret Taylor (Chair), Larry Summers, and Adam D'Angelo**.\"\n\n## Document type\n**Text messages, screenshot transcript.** Two pages of iMessage-style chat bubbles with timestamps. Murati's name on the left (\"Mira Murati\"), counterparty on the right (name redacted but addressed by Murati as \"**Kevin**\" — likely Microsoft CTO Kevin Scott). Date range Friday Nov 17, 2023 – Tuesday Nov 21, 2023. Bates MSFT_MUSK000056765–56766. Marked HIGHLY CONFIDENTIAL.\n\n## Logistics\n- **Trial admission:** Day 8 (May 6, 2026 — Murati live; Zilis live; Toner video begins). No prior wiki citation in the chunk metadata; admission day inferred from Box upload and Murati's live testimony that day.\n- **Box upload:** 2026-05-06 14:07:57 PT — Day 8 early-afternoon batch.\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~812 KB, 2 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `309.pdf`. Bates is `MSFT_MUSK000056765` — i.e., produced from Microsoft.\n\n## Transcribed text\n\n> **Friday, November 17, 2023**\n>\n> **Mira Murati** [openai] — 11:44:21 AM\n> Hi Kevin, this is Mira (openai). I have something extremely urgent and important. May I call you for 5 mins?\n>\n> **Mira Murati** — 7:32:16 PM\n> Kevin feel free to call anytime if you'd like to chat about any of this insanity tonight\n>\n> **Monday, November 20, 2023**\n>\n> **[Kevin]** — 4:07:29 AM\n> Sorry I missed your call. I'm in DC and just grabbed 3.5 hours of shuteye from 3a to 6:30 ET. Up now and can chat whenever.\n>\n> **Mira Murati** — 4:08:51 AM\n> Don't worry Kevin, we resolved. Close to having the board resign\n>\n> **[Kevin]** — 4:09:08 AM\n> For real this time?\n>\n> **Mira Murati** — 4:09:20 AM\n> It seems so.\n>\n> **Mira Murati** — 4:09:25 AM\n> Ilya signed our petition\n>\n> **[Kevin]** — 4:12:19 AM\n> I knew that you all had sent a petition. Can you send me what it said? Shockingly given how close everyone is covering this right now, I haven't heard mention of it, other than through OAI->Microsoft employee conversations.\n>\n> **[Kevin]** — 4:12:28 AM\n> And LMK if there's anything I can do to help you.\n>\n> **[Kevin]** — 4:12:34 AM\n> You must be exhausted.\n>\n> **Mira Murati** — 4:13:41 AM\n> [image]\n>\n> **Mira Murati** — 4:13:41 AM\n> Nearly 500 signatures. I think it hasn't leaked yet because we just put it together and will release it soon if the board doesn't resign immediately\n>\n> **Mira Murati** — 4:13:59 AM\n> Loved \"And LMK if there's anything I can do to help you.\"\n>\n> **[Kevin]** — 4:14:10 AM\n> It's been insane\n>\n> **Tuesday, November 21, 2023**\n>\n> **Mira Murati** — 9:33:08 PM\n> We've reached agreement. Sam will return as CEO with a new initial board Bret Taylor (Chair), Larry Summers, and Adam D'Angelo. We will complete details in next day but we're announcing to the company right away\n\n## Commentary\n\nPX 309 is the contemporaneous **inside-the-firing** text record from the OpenAI side of the November 2023 weekend. It corroborates Murati's Day 8 live testimony that \"**Microsoft and Satya in particular were important in unifying the team**\" *(5/6/2026 Testimony @ ~10:13 PT)* — see [[Key Themes]] § \"Microsoft's role in Altman's reinstatement.\" Plaintiffs use this for their \"[[Key Themes|captured by Microsoft]]\" theme: while the OpenAI nonprofit board was attempting to exercise its mission-protection authority by removing Altman, Murati was in real-time coordination with a Microsoft executive (\"Kevin\"), and a Microsoft-organized petition with \"**nearly 500 signatures**\" was the structural lever used to force the board's resignation. The chronology — Friday \"extremely urgent and important\" → Monday \"**Close to having the board resign**\" → Tuesday \"**We've reached agreement. Sam will return as CEO**\" — is plaintiffs' textual proof that the for-profit/Microsoft architecture has *practical* power to override the nonprofit board's mission-protection function (cf. Toner: \"**It created a threat**\" — [[Key Themes]] § \"Microsoft's role\"). The redacted Microsoft handle — addressed as \"Kevin\" — is consistent with **Kevin Scott**, Microsoft's CTO. See [[Day 8|Day 8 digest]] for Murati's full testimony arc.\n\n---\n*See also:* [[Day 8|Day 8 digest]] · [[Mira Murati]] · [[Sam Altman firing (Nov 2023)]] · [[Microsoft]] · [[Helen Toner]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-315", "exhibit": "PX 315", "party": "Plaintiffs", "type": "Text messages", "admitted_trial_day": "Day 8 (May 6, 2026)", "uploaded_box_pt": "2026-05-06T14:07:58", "uploader": "Someone", "pages": 7, "size_bytes": 1705226, "source_pdf": "PX-315.pdf", "pdf_url": "https://media.mts-in.com/PX-315.pdf", "body_markdown": "# PX 315 — November 19–20, 2023 Altman–Murati text messages during the OpenAI board firing\n\n> Cellebrite extract of 78 iMessage messages between Sam Altman and Mira Murati spanning the night Altman was fired by the OpenAI nonprofit board — Murati relaying the board's verdict in real time (\"Yes for you to be gone\"; \"They want new ceo in place\"; \"New guy is rando twitch guy\"; \"Just not your hand on agi\") while Altman tries to negotiate his return.\n\n## Document type\n**Text messages (forensic extract).** Cellebrite-style \"Short Message Report,\" 1 conversation, 3 listed participants (two Sam Altman entries reflecting different identifiers + Mira Murati), 78 total messages, date range 11/19/2023 – 11/20/2023. iMessage. Includes one inline iPhone-lock-screen screenshot (image attachment IMG_5924.PNG) showing notifications from Ashlee Vance and Ron Conway. Bates OPENAI_MUSK00027451 through OPENAI_MUSK00027457. Marked CONFIDENTIAL.\n\n## Logistics\n- **Trial admission:** Day 8 (May 6, 2026) — Murati live; Zilis live; Toner video begins.\n- **Box upload:** 2026-05-06 14:07:58 PT.\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~1.63 MB, 7 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `315.pdf`.\n\n## Transcribed text\n\n> **Short Message Report**\n> Conversations: 1 / Participants: 3 / Total Messages: 78 / Date Range: 11/19/2023 – 11/20/2023\n\n**Messages in chronological order (times shown in GMT +00:00):**\n\n> **SA — Sam Altman** — 11/19/2023, 5:43 PM\n> can you please officially invite me to the office for a meeting?\n>\n> **MM — mira murati** — 11/19/2023, 5:45 PM\n> Yes I will, do you have an update you can share?\n>\n> **SA — Sam Altman** — 11/19/2023, 5:52 PM\n> adam is trying to get the board to agree to a configuration\n>\n> **SA — Sam Altman** — 11/19/2023, 5:52 PM\n> he is now saying they need til end of day\n>\n> **SA — Sam Altman** — 11/19/2023, 5:52 PM\n> satya and i said that doesn't work\n>\n> **SA — Sam Altman** — 11/19/2023, 5:53 PM\n> and that we need to start preparing for plan b\n>\n> **MM — mira murati** — 11/19/2023, 5:54 PM\n> Ok please give me a second, I'm about to speak with them\n>\n> **SA — Sam Altman** — 11/19/2023, 5:54 PM\n> ok great\n>\n> **SA — Sam Altman** — 11/19/2023, 6:15 PM\n> have an update lmk when you can talk\n>\n> **SA — Sam Altman** — 11/20/2023, 1:37 AM\n> are you on with them? unrelated problem if you're still waiting\n>\n> **MM — mira murati** — 11/20/2023, 1:38 AM\n> Not yet, just in a quiet room because I didn't want all the outside theater\n>\n> **SA — Sam Altman** — 11/20/2023, 2:23 AM\n> can you indicate directionally good or bad? satya and others anxious\n>\n> **MM — mira murati** — 11/20/2023, 3:05 AM\n> Directionally very bad\n>\n> **SA — Sam Altman** — 11/20/2023, 3:05 AM\n> ok\n>\n> **SA — Sam Altman** — 11/20/2023, 3:06 AM\n> can you wrap up soon? lots of pressure from msft for an update\n>\n> **MM — mira murati** — 11/20/2023, 3:07 AM\n> Sam this is very bad\n>\n> **SA — Sam Altman** — 11/20/2023, 3:07 AM\n> can i come in?\n>\n> **MM — mira murati** — 11/20/2023, 3:07 AM\n> They don't want you to\n>\n> **SA — Sam Altman** — 11/20/2023, 3:08 AM\n> what do you want to make it better? i'm still willing to just walk away if that helps\n>\n> **SA — Sam Altman** — 11/20/2023, 3:08 AM\n> if they are ramped up for crazy lawsuits against me then i'm not sure what\n>\n> **SA — Sam Altman** — 11/20/2023, 3:09 AM\n> can you please tell them i just want to resolve this however and would like to join\n>\n> **MM — mira murati** — 11/20/2023, 3:11 AM\n> They're convinced about their decision\n>\n> **SA — Sam Altman** — 11/20/2023, 3:11 AM\n> for me to be fired? or some new thing?\n>\n> **MM — mira murati** — 11/20/2023, 3:11 AM\n> **Yes for you to be gone**\n>\n> **SA — Sam Altman** — 11/20/2023, 3:11 AM\n> ok\n>\n> **SA — Sam Altman** — 11/20/2023, 3:12 AM\n> then can i come in and talk about a path forward with them?\n>\n> **MM — mira murati** — 11/20/2023, 3:12 AM\n> They're saying no and they need more time\n>\n> **SA — Sam Altman** — 11/20/2023, 3:13 AM\n> more time for what?\n>\n> **MM — mira murati** — 11/20/2023, 3:17 AM\n> **They've walked me through all the reasons and the issues with you and why you can't be ceo**\n>\n> **SA — Sam Altman** — 11/20/2023, 3:18 AM\n> can you ask why they've been a saying all weekend they wanted me back?\n>\n> **MM — mira murati** — 11/20/2023, 3:18 AM\n> They want new ceo in place\n>\n> **SA — Sam Altman** — 11/20/2023, 3:20 AM\n> can you say you will call back in 10 min\n>\n> **MM — mira murati** — 11/20/2023, 3:24 AM\n> **They want to have a new ceo in place tonight (not me**\n>\n> **SA — Sam Altman** — 11/20/2023, 3:26 AM\n> do they know who?\n>\n> **SA — Sam Altman** — 11/20/2023, 3:26 AM\n> can i tell satya? is this final?\n>\n> **SA — Sam Altman** — 11/20/2023, 3:29 AM\n> or, should you add satya in?\n>\n> **MM — mira murati** — 11/20/2023, 3:29 AM\n> Trying to add Satya now\n>\n> **SA — Sam Altman** — 11/20/2023, 3:29 AM\n> still don't want me?\n>\n> **MM — mira murati** — 11/20/2023, 3:29 AM\n> **New guy is rando twitch guy**\n>\n> **MM — mira murati** — 11/20/2023, 3:29 AM\n> They don't want you\n>\n> **SA — Sam Altman** — 11/20/2023, 3:29 AM\n> emmett?\n>\n> **MM — mira murati** — 11/20/2023, 3:30 AM\n> Yeah\n>\n> **MM — mira murati** — 11/20/2023, 3:30 AM\n> But hold on I'm pulling Satya now\n>\n> **SA — Sam Altman** — 11/20/2023, 3:30 AM\n> ok\n>\n> **SA — Sam Altman** — 11/20/2023, 3:31 AM\n> do you think any way you can turn this around? even if we let them sleep on it tonight or whatever?\n>\n> **MM — mira murati** — 11/20/2023, 3:31 AM\n> Have been trying\n>\n> **SA — Sam Altman** — 11/20/2023, 3:37 AM\n> from a journalist:\n>\n> **SA — Sam Altman** — 11/20/2023, 3:38 AM\n> [Image attachment: iPhone lock screen, Sunday November 19, 7:37, showing Signal notifications from \"Ashlee Vance\" — \"i think board has lined up a ceo plan to announce\"; \"i dunno if this has come to you or not but passing along\" — and a Ron Conway iMessage to Brian Chesky and 2 others]\n>\n> **MM — mira murati** — 11/20/2023, 3:38 AM\n> Yeah\n>\n> **MM — mira murati** — 11/20/2023, 3:39 AM\n> Hoping Satya can help undo this\n>\n> **SA — Sam Altman** — 11/20/2023, 3:39 AM\n> should team send letter to board now?\n>\n> **SA — Sam Altman** — 11/20/2023, 3:39 AM\n> is he on with them?\n>\n> **MM — mira murati** — 11/20/2023, 3:39 AM\n> It won't matter\n>\n> **MM — mira murati** — 11/20/2023, 3:39 AM\n> They don't care if everyone quits\n>\n> **SA — Sam Altman** — 11/20/2023, 3:39 AM\n> **is that what they want is the IP going to anthropic?**\n>\n> **SA — Sam Altman** — 11/20/2023, 3:39 AM\n> that's what team thinks\n>\n> **MM — mira murati** — 11/20/2023, 3:39 AM\n> **Just not your hand on agi**\n>\n> **SA — Sam Altman** — 11/20/2023, 3:40 AM\n> i can not come back!\n>\n> **SA — Sam Altman** — 11/20/2023, 3:40 AM\n> they were the ones that asked as of yesterday morning\n>\n> **SA — Sam Altman** — 11/20/2023, 3:40 AM\n> did satya get on call?\n>\n> **MM — mira murati** — 11/20/2023, 3:42 AM\n> Yes with him\n>\n> **SA — Sam Altman** — 11/20/2023, 3:50 AM\n> wait i have an interesting idea\n>\n> **MM — mira murati** — 11/20/2023, 3:50 AM\n> Still with Satya\n>\n> **MM — mira murati** — 11/20/2023, 3:50 AM\n> Go ahead\n>\n> **SA — Sam Altman** — 11/20/2023, 3:50 AM\n> **what if msft acquires openai? would that provide the governance the board wants?**\n>\n> **SA — Sam Altman** — 11/20/2023, 3:51 AM\n> is satya making progress at all\n>\n> **MM — mira murati** — 11/20/2023, 3:51 AM\n> Satya is being diplomatic\n>\n> **SA — Sam Altman** — 11/20/2023, 4:36 AM\n> will you emmett on with the team asap\n>\n> **SA — Sam Altman** — 11/20/2023, 4:36 AM\n> also does adam know you rehired me\n>\n> **MM — mira murati** — 11/20/2023, 4:36 AM\n> Yes\n>\n> **SA — Sam Altman** — 11/20/2023, 6:10 AM\n> can you chat soon?\n>\n> **MM — mira murati** — 11/20/2023, 7:47 AM\n> Thought about your proposal, I'm on board\n>\n> **SA — Sam Altman** — 11/20/2023, 7:50 AM\n> kk. i think it's best shot to get company back and keep people\n>\n> **SA — Sam Altman** — 11/20/2023, 7:50 AM\n> and if that doesn't work\n>\n> **SA — Sam Altman** — 11/20/2023, 7:51 AM\n> we have a new thing and the compute and the ip\n>\n> **SA — Sam Altman** — 11/20/2023, 7:51 AM\n> but i feel has good chance of winning\n>\n> **SA — Sam Altman** — 11/20/2023, 7:51 AM\n> i think you all just need to get a petition of everyone saying they will quit and join, they don't even do\n>\n> **SA — Sam Altman** — 11/20/2023, 7:53 AM\n> lmk if you have a minute for a call?\n\n## Commentary\n\nPX 315 is the **interior view** of the November 2023 OpenAI board firing — the moment that runs through [[Day 8|Day 8]] \"like a current\" per [[Key Themes]]. The thread captures Murati relaying the board's reasoning in near-real time (\"**they've walked me through all the reasons and the issues with you and why you can't be ceo**\"; \"**Just not your hand on agi**\") and Altman's escalating proposals: walk away, return on different terms, and finally the extraordinary \"**what if msft acquires openai? would that provide the governance the board wants?**\" Plaintiffs use this exhibit (and Murati's live testimony — \"Was [Mr. Altman] candid with you? — Not always\") as evidence that the board's stated rationale was *Altman's relationship with the truth* rather than mere strategic disagreement, and that the for-profit/Microsoft architecture has [[Key Themes|practical power to override the nonprofit board's mission-protection function]] — Helen Toner's video frames Microsoft's hire-everyone offer as \"**a threat**\" to the board's ability to carry out its mission. The \"is the IP going to anthropic?\" line is the unprompted Altman fear that the board would simply transfer the technology elsewhere — and \"**not your hand on agi**\" is the board's own articulated objection in Murati's voice. The Ashlee Vance Signal screenshot — incoming press leaks within hours of Altman's firing — is the contemporaneous record of how fast the public-facing crisis ran. See [[Day 8|Day 8 digest]] for Murati's live testimony and Toner's video deposition; see also [[Key Themes]] § \"Day 8 — Sam Altman's candor.\"\n\n---\n*See also:* [[Day 8|Day 8 digest]] · [[Mira Murati]] · [[Sam Altman]] · [[Helen Toner]] · [[Microsoft Thread]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-319", "exhibit": "PX 319", "party": "Plaintiffs", "type": "Slack messages", "admitted_trial_day": "unknown", "pages": 2, "size_bytes": 1985316, "source_pdf": "PX-319.pdf", "bates": "SUTSKEVER_MUSKSUB_00000525-526", "also_marked": "DX 1049", "pdf_url": "https://media.mts-in.com/PX-319.pdf", "body_markdown": "# PX 319 — Helen Toner @channel Slack message announcing Emmett Shear as interim CEO (Nov 20, 2023)\n\n> **The plaintiffs' marking of the same document as [[DX 1049]]** — the board's all-hands Slack message sent **Mon 11/20/2023 at 1:42 AM UTC** declaring \"**Sam will not return as CEO**\" and announcing **Emmett Shear** as interim. Identical Bates (`SUTSKEVER_MUSKSUB_00000525–526`); both sides relied on the same message.\n\n## Document type\n**Slack message.** From: **Helen Toner**. To: **Ilya Sutskever**. Cc: **Adam D'Angelo, Tasha McCauley**. Marked **HIGHLY CONFIDENTIAL — ATTORNEYS' EYES ONLY**.\n\n## Logistics\n- **Trial admission:** unknown.\n- **File size:** ~2.0 MB, 2 pages. A second source PDF (`0319.pdf` ≈ 2.0 MB, also Bates 319.002) exists alongside; both are scans of the same content. Stored at `Elon Musk, et al. v. Samuel Altman, et al/PX-319.pdf` and `PX-319-alt.pdf`.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `319.pdf` (and `0319.pdf`).\n- **Cross-marked:** Identical content (and Bates) to **[[DX 1049]]**.\n\n## Transcribed text\n\nSee **[[DX 1049]]** for the full transcribed text — plaintiffs' PX 319 marking is of the **same** Helen Toner @channel Slack message (Bates `SUTSKEVER_MUSKSUB_00000525–526`).\n\nThe pull-quotes most relevant to plaintiffs' theory:\n\n> \"**Sam's behavior and lack of transparency in his interactions with the board undermined the board's ability to effectively supervise the company in the manner it was mandated to do.**\"\n>\n> \"**Despite rumors to the contrary, Sam will not return as CEO.**\"\n>\n> \"OpenAI was deliberately structured as a non-profit entity governing the **capped-profit entity**. The board was purposefully established with the intention of **independent oversight** to advance OpenAI's mission.\"\n\n## Commentary\n\nThe exhibit is **dual-marked** by both sides — a strong signal that the message is anchor-quality on both theories. Plaintiffs use it to frame the candor problem as the board itself characterized it, in writing, on the firing weekend; defense uses the same message's \"**not about product safety or security … not about any singular incident … governance issue**\" framing to keep the firing narrative away from the safety / commercialization framing plaintiffs pursue. See [[DX 1049]] for full transcription, commentary, and cross-references.\n\n---\n*See also:* [[DX 1049]] · [[DX 1040]] · [[Day 8]] · [[Day 9]] · [[Helen Toner]] · [[Tasha McCauley]] · [[Adam D'Angelo]] · [[Ilya Sutskever]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-350", "exhibit": "PX 350", "party": "Plaintiffs", "type": "Email thread", "admitted_trial_day": "Day 6 (May 4, 2026)", "uploaded_box_pt": "2026-05-04T14:46:02", "uploader": "Someone", "pages": 2, "size_bytes": 280267, "source_pdf": "PX-350.pdf", "pdf_url": "https://media.mts-in.com/PX-350.pdf", "body_markdown": "# PX 350 — Aug 18, 2017 Birchall / Musk / Brockman email thread on Altman's family-office equity grant to Brockman\n\n> The email thread that put the **2017 ~$10M Altman family-office equity grant to Greg Brockman** in front of Musk for the first time — Birchall reports it to Musk; Musk forwards to Brockman with two question marks (\"??\"); Brockman replies (and forwards the whole exchange to Sutskever) acknowledging the 1% grant in Sam's family office, valued at \"**O($10M)**,\" in lieu of the YC stock that had run out.\n\n## Document type\n**Email thread, 2 pages.** Inner messages: (1) Jared Birchall <jared@excession.com> to Elon Musk, Aug 18, 2017 8:26:45 AM PDT, \"Re: OpenAI\"; (2) Musk forwards to Brockman with \"??\", Aug 18, 2017 8:31 AM; (3) Brockman replies to Musk, Aug 18, 2017 9:42 AM; (4) Brockman forwards the whole chain to Ilya Sutskever, Aug 18, 2017 5:29:31 PM. Bates OPENAI_MUSK00023430–00023431. Stamped CONFIDENTIAL.\n\n## Logistics\n- **Trial admission:** Day 6 (May 4, 2026) — Russell direct + cross; Brockman cross by Mr. Kry. Per the wiki TSV reference: \"**PX 350 — 2017 Altman family-office equity grant to Brockman.**\" Used by Mr. Kry on Brockman cross — see [[Key Themes]] §\"Greg Brockman's contributions ('zero')\": \"The hidden **$10 million** family-office equity grant from Sam Altman in 2017 (PX 350) that Brockman never disclosed to Musk; Musk forwarded the discovery to Brockman with '??'.\" (`Day 6 @ ~10:08 PT`.)\n- **Box upload:** 2026-05-04 14:46:02 PT — Day 6 mid-afternoon batch.\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~274 KB, 2 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `350.pdf`.\n\n## Transcribed text\n\n> **From:** Greg Brockman <gdb@openai.com>\n> **To:** Ilya Sutskever <ilyasu@openai.com>\n> **Subject:** Fwd: OpenAI\n> **Date:** 8/18/2017 5:29:31 PM\n>\n> ---------- Forwarded message ----------\n> **From:** Greg Brockman <gdb@openai.com>\n> **To:** Elon Musk <erm@spacex.com>\n> **Subject:** Re: OpenAI\n> **Date:** Fri, Aug 18, 2017 at 9:42 AM\n>\n> My original offer was: $175k salary, 50bps of YC stock, and 50 bps of YC continuity. But we ran out of YC stock fulfilling others' offers, and Sam gave me a 1% grant in his family office instead. There's no personal loyalty to Sam here — we ran out of OpenAI's YC stock as we hadn't tracked the allocation closely enough while making offers, and this was fulfilling my original offer. I feel equal loyalty to both of you.\n>\n> Even if valued at O($10M), it does not materially affect my personal financial situation. And obviously it could end up being worth less.\n>\n> I'm not that motivated by money, but I am motivated by fairness, and it would have felt bad for me to get nothing after the YC stock ran out. FWIW I *am* motivated by public recognition for my own work, as I mentioned before the Dota announcement.\n>\n> - gdb\n\n> **From:** Elon Musk <erm@spacex.com>\n> **To:** Greg Brockman <gdb@openai.com>\n> **Subject:** Re: OpenAI\n> **Date:** Fri, Aug 18, 2017 at 8:31 AM\n>\n> ??\n\n> **From:** Jared Birchall <jared@excession.com>\n> **To:** Elon Musk <erm@spacex.com>\n> **Subject:** Re: OpenAI\n> **Date:** August 18, 2017 at 8:26:45 AM PDT\n>\n> I had a fairly extensive conversation with Sam Altman yesterday in an attempt to understand his thinking on the numbers. I can share greater detail when we meet today, but one thing worth mentioning now is that he compensated Greg on the side, giving him a percentage ownership of the LLC that holds the assets of Sam's personal family office — a stake which Sam says equates to ~$10M. **Naturally Greg is going to have a greater allegiance toward Sam as a result of this arrangement.**\n>\n> --\n> Sent from mobile\n\n## Commentary\n\nPX 350 is the documentary spine of plaintiffs' \"**Greg Brockman's contributions: zero**\" theme — the \"hidden $10 million\" Altman side grant that defense will cast as backfill for an unfulfilled YC equity offer and plaintiffs cast as the moment Brockman's loyalties became financially aligned to Altman, not to the OpenAI nonprofit (see [[Key Themes#Greg Brockman's contributions (\"zero\")|Key Themes]]). Three things land in the text itself: (1) **Birchall's contemporaneous read** — \"Naturally Greg is going to have a greater allegiance toward Sam as a result of this arrangement\" — written by the man Birchall, who later told the jury Brockman's contributions were \"**zero**\" (`043026TT.txt:4416`); (2) **Musk's two question marks** — the only thing he wrote — used on Brockman's Day 6 cross to land plaintiffs' point that Musk learned about the grant only because his family-office head dug it out, not because Brockman had told him; and (3) **Brockman's own valuation of the grant**: \"Even if valued at O($10M), it does not materially affect my personal financial situation\" — a 2017 written admission that arrived in front of the jury alongside his Day 6 testimony that he later acquired a **~$30B** OpenAI for-profit stake having paid **$0** for it. The thread also forms the contemporaneous backdrop for the Aug. 21, 2017 [[Brockman Journal]] entry \"**This is the only chance we have to get out from under Elon**\" three days later (PX 151 / [[Key Themes]]), and for the September 2017 \"Honest Thoughts\" / [[PX-157|\"final straw\"]] negotiation where the Cerebras and structure questions surfaced in the open.\n\n---\n*See also:* [[Day 6|Day 6 digest]] · [[Greg Brockman]] · [[Sam Altman]] · [[Brockman Journal]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-355", "exhibit": "PX 355", "party": "Plaintiffs", "type": "Text messages", "admitted_trial_day": "Day 3 (April 29, 2026)", "uploaded_box_pt": "2026-04-29T15:09:18", "uploader": "Someone", "pages": 3, "size_bytes": 182136, "source_pdf": "PX-355.pdf", "pdf_url": "https://media.mts-in.com/PX-355.pdf", "body_markdown": "# PX 355 — May 1, 2023 Musk/Spiro/Altman texts re: Microsoft compute deal & \"de facto control of AGI\"\n\n> A 14-message group text on May 1, 2023 between Elon Musk, Sam Altman, Alex Spiro, and Jared Birchall — Spiro asks for copies of the OpenAI structure documents, Altman pushes lawyers'-eyes-only on-site review, and Musk states his core concern: that Microsoft \"**will have de facto control of AGI**.\"\n\n## Document type\n**Text messages, Cellebrite-style \"Short Message Report\" extraction.** Three pages: cover sheet (1 conversation `chat921307012919564707`, 14 messages, 5 participants — Elon Musk, Sam Altman, Alex Spiro, Jared Birchall, plus a fifth participant whose handle is fully redacted) and two pages of message-by-message transcript with timestamps in GMT-07:00. Bates 2024MUSK-0006349–6351.\n\n## Logistics\n- **Trial admission:** Day 3 (April 29, 2026). Used during Musk's direct/cross to anchor the May 2023 timing of Musk's \"phase three\" disillusionment over the Microsoft relationship (see [[Key Themes]] § \"Phase three — 2023 Microsoft control concerns\").\n- **Box upload:** 2026-04-29 15:09:18 PT — Day 3 mid-afternoon batch (clustered with PX 157, PX 233, PX 296).\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~178 KB, 3 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `355.pdf`.\n\n## Transcribed text\n\n> **Conversation:** chat921307012919564707\n> **Date Range:** 5/1/2023\n> **Participants:** Elon Musk, Sam Altman, Alex Spiro, Jared Birchall, [one redacted]\n>\n> **Messages in chronological order (GMT -07:00):**\n>\n> **EM** (Elon Musk) — 5/1/2023, 12:15 PM\n> Can Alex show up tomorrow?\n>\n> **SA** (Sam Altman) — 5/1/2023, 1:27 PM\n> yes tomorrow works. to confirm my understanding: alex (and jared if he wants) will show up tomorrow, and we will make available to review on-site all structure docs + all docs about control + relationship with microsoft. our GC jason kwon will be available to answer questions. you all will keep this confidential to yourselves, and alex offered that we can mark things that might be competitive intel wrt to x as lawyers eyes only. sound good?\n>\n> **AS** (Alex Spiro) — 5/1/2023, 1:29 PM\n> We need copies.\n>\n> **AS** — 5/1/2023, 1:30 PM\n> Not merely review on site. That's not feasible.\n>\n> **EM** — 5/1/2023, 1:37 PM\n> Liked \"yes tomorrow works. to confirm my understanding: a…\"\n>\n> **EM** — 5/1/2023, 1:37 PM\n> It's ok\n>\n> **EM** — 5/1/2023, 1:44 PM\n> I doubt that there's anything that's competitive intel. The point is to understand the relationship between all the companies and the original OpenAI 501c3.\n>\n> It would be absurd to agree not to use the information learned if what we learn seems improper.\n>\n> **SA** — 5/1/2023, 1:53 PM\n> oh on all of that of course; fully aligned and of course you can use info you think is improper\n>\n> **SA** — 5/1/2023, 1:54 PM\n> i meant, for example, the specifics of our compute deal with microsoft. is that something you're interested in?\n>\n> **SA** — 5/1/2023, 1:54 PM\n> if it's just the relationship between the companies and the original 501c3 that's easy and straightforward\n>\n> **EM** — 5/1/2023, 1:59 PM\n> For sure, understanding what rights Microsoft has is important. **One of the things I'm concerned about is that they will have de facto control of AGI.**\n>\n> **SA** — 5/1/2023, 2:35 PM\n> ok we can make this even simpler for you guys. we will just send you electronic read-only versions of the structure docs. those also contain key MS rights and control stuff you are interested in.\n>\n> then we can also share the MS compute deal docs in a way that gives you reassurance on AGI. alex can come view them as lawyers eyes only docs because they are <100 pages total.\n>\n> sound ok?\n>\n> **EM** — 5/1/2023, 2:39 PM\n> Probably ok\n>\n> **SA** — 5/1/2023, 2:40 PM\n> cool\n\n## Commentary\n\nPX 355 is plaintiffs' best on-the-record statement of Musk's actual contemporaneous concern in spring 2023 — not \"stole-a-charity\" but \"**de facto control of AGI**\" by Microsoft — and it puts that concern in Musk's mouth roughly ten months before he filed suit. The exchange also catches Altman in a cooperative posture — offering to make structure documents available for review and to share the Microsoft compute deal \"in a way that gives you reassurance on AGI\" — which the defense uses on cross to argue that Musk had every opportunity to inspect the disputed structure and chose to litigate instead. Spiro's role here matters: by May 2023, Musk's litigation counsel was already in the loop and demanding copies, foreshadowing the February 2024 complaint. The texting context — Musk in possession of his own large language model effort at xAI (announced July 2023, ~two months later) — supports the defense's [[Key Themes]] § \"Tesla / xAI as Musk's own AGI play\" argument. Note: the fifth, fully redacted participant on the cover sheet is unidentified in the transcript and may be a paralegal or assistant added for delivery.\n\n---\n*See also:* [[Day 3|Day 3 digest]] · [[Sam Altman]] · [[Alex Spiro]] · [[Jared Birchall]] · [[Phases One Two Three]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-359", "exhibit": "PX 359", "party": "Plaintiffs", "type": "Email + attachment (charter draft)", "admitted_trial_day": "Day 6 (May 4, 2026)", "uploaded_box_pt": "2026-05-04T14:46:02", "uploader": "Someone", "pages": 2, "size_bytes": 433327, "source_pdf": "PX-359.pdf", "pdf_url": "https://media.mts-in.com/PX-359.pdf", "body_markdown": "# PX 359 — Dec 28–30, 2023 Brockman → Summers/Kwon \"Charter 2.0\" draft (with Larry Summers benefit-corp suggestion)\n\n> Greg Brockman's December 30, 2023 internal forward of his \"Charter 2.0\" draft to Jason Kwon and Sam Altman, atop a December 28, 2023 email exchange with Larry Summers about converting OpenAI to a benefit corporation — the document that retires the original \"unconstrained by a need to generate financial return\" framing in favor of \"capitalism… as a positive force.\"\n\n## Document type\n**Email thread + attached draft text, two pages.** Brockman's December 30, 2023 5:07 AM internal forward to Kwon and Altman, atop his December 28, 2023 10:03 AM reply to Larry Summers (cc Altman, Bret Taylor, Adam D'Angelo) embedding the \"Charter 2.0\" draft, atop Summers' December 28, 2023 7:24 AM email noting \"musk's thing is like anthropic is a benefit corp.\" Subject \"Fwd: Benefit corp.\" Marked \"[privileged]\" at top. Bates OPENAI_MUSK00030769–30770.\n\n## Logistics\n- **Trial admission:** Day 6 (May 4, 2026). Pre-trial wiki reference: **\"PX 359 — Charter 2.0 (Dec 2023).\"** Used during Brockman cross by Mr. Kry to anchor the December 2023 board-rebuild moment (post-Altman-firing reinstatement) and to contrast with the [[PX 24]] 2018 Charter language.\n- **Box upload:** 2026-05-04 14:46:02 PT — Day 6 early-afternoon batch (clustered with [[PX 90]]).\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~423 KB, 2 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `359.pdf`.\n\n## Transcribed text\n\n> **From:** Greg Brockman <gdb@openai.com>\n> **To:** Jason Kwon <jason@openai.com>, Sam Altman <sama@openai.com>\n> **Subject:** Fwd: Benefit corp\n> **Date:** 12/30/2023 5:07:13 AM\n>\n> [privileged]\n\n> **From:** Greg Brockman <gdb@openai.com>\n> **To:** Larry Summers <lhs@openai.com>\n> **Cc:** Sam Altman <sama@openai.com>, Bret Taylor <btaylor@openai.com>, Adam D'Angelo <ad@openai.com>\n> **Subject:** Re: Benefit corp\n> **Date:** Thu, Dec 28, 2023 at 10:03 AM\n>\n> Great to hear, I am also intrigued by benefit corp. Annual report within the current structure could be interesting!\n>\n> As part of that, would definitely like to align on how we think about the mission. Sam and I have been working on Charter 2.0 as mentioned at the board meeting. Here's our current draft, would love any early feedback on the direction:\n>\n> ---\n>\n> **Charter 2.0**\n>\n> We started OpenAI with the goal \"to advance digital intelligence in the way that is most likely to benefit humanity as a whole\"; this remains our fundamental guiding principle.\n>\n> Over the years, we've increasingly learned how to further this goal in practice, such as by making breakthroughs and deploying them as products like ChatGPT. We've also realized we were wrong at times. **For example, we originally set the company up as a non-profit \"unconstrained by a need to generate financial return\"; over the past eight years we've grown to regard capitalism not as a constraint but instead as a positive force that, managed well, can align different stakeholders to pool their resources to create far more value than they keep.**\n>\n> While we know we'll make changes as the technology continues to unfold, the following is our articulation of how we pursue our fundamental goal today:\n>\n> **OpenAI's mission is to provide AGI that benefits all of humanity.**\n>\n> **Empower humanity**\n>\n> We provide AI tools that allow others to shape the future. We have deep conviction that much more good than bad will come of this, since we believe:\n>\n> (1) AI can benefit every aspect of our lives — from helping with daily tasks to one day helping humanity solve our largest challenges.\n> (2) The collective effort of humanity will be required to figure out how to best apply AI, and\n> (3) AI's downsides can be mitigated by collectively working on topics such as technical alignment, trading off misuse mitigations with liberty, maximizing race-to-the-top and minimizing race-to-the-bottom dynamics on safety, predicting AI capabilities, responding to societal impacts, etc..\n>\n> **Mission success requires that humanity as a whole, our customers, and our shareholders all win.** That is, that everyone's quality of life is radically better (imagine a world where everyone can figure out how to be happier and more fulfilled than anyone is today), our customers are accomplishing things they couldn't have dreamed of otherwise, and our employees and investors are making a return commensurate with making all of this possible.\n>\n> We focus on producing the most capable models even if they require too much compute to serve at humanity scale today. We are working in parallel to scale up the world's supply of compute, which we expect to become a critical limited resource in the future.\n>\n> **Deploy iteratively** [[better name needed here: meaning is \"don't let an overhang of capability build up\" or \"keep changes in the world continuous to match the continuity of technological progress\".]]\n>\n> We choose projects in order to maximize our rate of learning about and preparation for AGI. This generally means our top priority is our research, but we also deploy useful services, build large-scale inference infrastructure, work with governments and the public on understanding and responding to the upsides and downsides of this technology, explore new methods for getting democratic input on what an AI should do, etc..\n>\n> We believe the mission is more likely to succeed when society can co-evolve with AI advances, so we work to iteratively deploy our technology broadly and are excited to see an emerging constellation of AI efforts with a broad range of conflicting perspectives. Since next year's AI will be much better than today's, we see risks increasing the longer a given system exists only in a lab or is known only to a select few.\n>\n> We regard building an AGI-level system as our first milestone. We plan for our activities to continue uninterrupted — though with increasing global coordination — into a world with a spectrum of AGIs and as humanity contemplates approaching superintelligence.\n>\n> - gdb\n\n> **From:** Larry Summers <lhs@openai.com>\n> **To:** Greg Brockman <gdb@openai.com>\n> **Cc:** Sam Altman <sama@openai.com>, Bret Taylor <btaylor@openai.com>, Adam D'Angelo <ad@openai.com>\n> **Subject:** Benefit corp\n> **Date:** Thu, Dec 28, 2023 at 7:24 AM\n>\n> I note that musk's thing is like anthropic is a benefit corp.\n> >\n> > I think we are agreed that looking into this or things related to it should be an early priority for the new board.\n> >\n> > Reading a bit about benefit corps it appears that one salient aspect involves an annual report on meeting public responsibilities and the like. It occurs to me that within our current structure we could prepare and issue such a report. It might preserve flexibility to do such a thing and it would not be much of a lift.\n> >\n> > Sent from my iPad\n\n## Commentary\n\nPX 359 is plaintiffs' \"phase three\" smoking gun — drafted in the six weeks **after** the November 2023 Altman firing/reinstatement. It is the textual moment OpenAI's leadership puts the original [[PX 24]] mission language (\"**unconstrained by a need to generate financial return**\") in the past tense and reframes capitalism as \"a positive force.\" Plaintiffs argue this draft is the concealed predicate for the eventual public for-profit/benefit-corp conversion (announced 2024–2025). Larry Summers' opening line — \"**musk's thing is like anthropic is a benefit corp**\" — is itself read by plaintiffs as evidence that the new board treated Musk's competing AI ventures as a reference architecture *while* litigating against Musk. The \"**humanity as a whole, our customers, and our shareholders all win**\" formulation is the language plaintiffs say replaced the original AGI-for-humanity charter promise. Cross-reference: see [[PX 24]] for the 2018 Charter; see [[Day 6|Day 6 digest]] for Brockman cross on the conversion timeline; see [[Key Themes#Musk's \"three phases\"|\"phase three\"]].\n\n---\n*See also:* [[Day 6|Day 6 digest]] · [[Greg Brockman]] · [[Sam Altman]] · [[PX 24]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-388", "exhibit": "PX 388", "party": "Plaintiffs", "type": "Demonstrative", "admitted_trial_day": "Day 2 (April 28, 2026)", "uploaded_box_pt": "2026-04-28T14:52:02", "uploader": "Someone", "pages": 1, "size_bytes": 763473, "source_pdf": "PX-388.pdf", "pdf_url": "https://media.mts-in.com/PX-388.pdf", "body_markdown": "# PX 388 — Photograph: Jensen Huang delivers first NVIDIA DGX-1 to OpenAI (Aug 2016)\n\n> A single photograph admitted as a demonstrative: NVIDIA CEO Jensen Huang in OpenAI's Pioneer Building offices, handing the company's first DGX-1 supercomputer to Greg Brockman, with Elon Musk standing in the background under the now-iconic hand-painted wall slogan \"MAN HAS A LARGE CAPACITY FOR EFFORT…\"\n\n## Document type\n**Demonstrative — single photograph.** Color image taken inside the OpenAI office, August 2016. Foreground: Jensen Huang (NVIDIA, CEO) handing a hardware unit to Greg Brockman; on a black coffee table is the silver/gold-trimmed NVIDIA DGX-1 chassis. Background: Elon Musk (arms crossed, dark jacket) and a third figure in profile; multiple staff in the foreground photographing the moment with phones. Bates 2024MUSK-0014490.\n\n## Logistics\n- **Trial admission:** Day 2 (April 28, 2026) — used during plaintiffs' opening / Musk direct as a visual anchor for OpenAI's nonprofit-era research culture and the Pioneer Building's importance to the case.\n- **Box upload:** 2026-04-28 14:52:02 PT.\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~746 KB, 1 page.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `388.pdf`.\n\n## Transcribed text\n\n> *[Color photograph, OpenAI Pioneer Building offices, August 2016.]*\n>\n> *Visible on the wall, partially obscured by figures, in hand-painted block letters:*\n>\n> > \"MAN HAS A LARGE CAPACITY FOR EFFORT. [...] THAN WE THINK IT IS THAT FEW EVER REACH [...] VALUE THE FACULTY OF KNOWING [WHAT WE...] THE [...] DO IT. KNOWING IS [...] IT IS [...] THE [C...] [I]SSUE IS NOT WHAT [...] WE KNO[W...] [...]EVE THAT IT IS [...] THE FAT[E OF THE] WORLD DEPE[NDS ON ...]\"\n>\n> *Foreground figures (left to right):* Jensen Huang (handing object); Greg Brockman (receiving); a young staffer (foreground left, partially turned).\n>\n> *Background figures:* Elon Musk (standing, arms crossed); an unidentified man in profile.\n>\n> *Object on table:* NVIDIA DGX-1 deep-learning supercomputer, in its production silver/black chassis.\n\n## Commentary\n\nPX 388 is the only photograph in plaintiffs' Day 2 opening — and it does work that words can't. It pictures the August 2016 delivery of NVIDIA's first DGX-1 to OpenAI, **personally hand-delivered by Jensen Huang** and signed \"*To the OpenAI team & the future of computing*\" — one of the most-photographed moments of the 2016 deep-learning boom. For plaintiffs the photo functions as a visual definition of the charity Musk says he funded: a small lab in a converted SF chocolate factory, with a hand-painted Teddy Roosevelt-flavored wall quote about effort and humility, receiving a piece of donated hardware (the DGX-1 was given to OpenAI by NVIDIA as a publicity gift). Musk's presence \"arms crossed\" in the background — not a featured donor, just one observer — undercuts defense's \"Musk wanted unilateral control\" theme. The image also documents the **Pioneer Building** as the operational seat of nonprofit-era OpenAI, making it a visual companion to the Pioneer-Building lease materials in [[Key Themes]] (DX 539, DX 600, [[PX-103]]). Cross-reference [[Greg Brockman]]'s testimony on the OpenAI early-research culture (Day 6).\n\n---\n*See also:* [[Day 2|Day 2 digest]] · [[Greg Brockman]] · [[Elon Musk]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-390", "exhibit": "PX 390", "party": "Plaintiffs", "type": "Court filing", "admitted_trial_day": "Day 6 (May 4, 2026)", "uploaded_box_pt": "2026-05-04T14:46:02", "uploader": "Someone", "pages": 10, "size_bytes": 330306, "source_pdf": "PX-390.pdf", "pdf_url": "https://media.mts-in.com/PX-390.pdf", "body_markdown": "# PX 390 — Greg Brockman's Fourth Amended Supplemental Responses to Musk's First Set of Interrogatories (Apr. 29, 2026)\n\n> The April 29, 2026 verified discovery response, signed by Brockman and his Wachtell/Morrison & Foerster team, **listing every entity in which Brockman holds a financial interest that has done business with OpenAI** — Cerebras, CoreWeave, Helion Energy, Stripe, and at least four further redacted entities — with date acquired, share count, FMV, and a corresponding narrative of OpenAI's transactions with each: a $11.945B March 2025 CoreWeave order, a 750-MW December 2025 Cerebras inference deal, the May 2024 / March 2026 Helion fusion-power agreements, and the $2.5M-then-$60M Stripe services-agreement chain.\n\n## Document type\n**Court filing — verified interrogatory response, 10 pages.** Caption: *Elon Musk, et al. v. Samuel Altman, et al.*, Case No. 4:24-cv-04722-YGR (N.D. Cal. Oakland Div.). Filed by Jordan Eth (Morrison & Foerster) and William Savitt et al. (Wachtell, Lipton, Rosen & Katz). Verified by Greg Brockman under penalty of perjury, 4/29/2026. Stamped \"Highly Confidential\" on every page; redactions extensive — entity-name lead-ins on at least four entities, share counts on six entities, all narrative paragraphs for two transactions blacked out.\n\n## Logistics\n- **Trial admission:** Day 6 (May 4, 2026). Used by Mr. Kry to walk through Brockman's web of business interests in OpenAI counterparties. The document underpins the [[Key Themes]] § \"Greg Brockman's contributions ('zero')\" Cerebras-conflict thread and the December 2025 ~$10B Cerebras chip deal that triggered the ~$8B-to-$23B Cerebras revaluation.\n- **Box upload:** 2026-05-04 14:46:02 PT — Day 6 mid-afternoon batch, the first file uploaded that afternoon (clustered with PX 151, PX 161, PX 241).\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~322 KB, 10 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `390.pdf`.\n\n## Transcribed text\n\n> **UNITED STATES DISTRICT COURT — NORTHERN DISTRICT OF CALIFORNIA — OAKLAND DIVISION**\n>\n> *ELON MUSK, et al., Plaintiffs, v. SAMUEL ALTMAN, et al., Defendants.* — Case No.: 4:24-cv-04722-YGR\n>\n> **GREGORY BROCKMAN'S FOURTH AMENDED SUPPLEMENTAL RESPONSES AND OBJECTIONS TO PLAINTIFF MUSK'S FIRST SET OF INTERROGATORIES**\n>\n> Pursuant to Rules 26 and 33 of the Federal Rules of Civil Procedure, Defendant Gregory Brockman by and through his attorneys, hereby amends his supplemental responses and objections to Plaintiff Musk's First Set of Interrogatories to Defendant Gregory Brockman, dated April 16, 2025… Brockman incorporates his responses and objections served on May 7, 2025, including his General Objections, Objections to Definitions and Instructions, and responses and objections to the specific interrogatories covered below, as though fully set forth herein.\n>\n> **INTERROGATORY NO. 4:** Identify all financial interests, direct or indirect, held by You in any entity that has conducted business with any OpenAI entity, Including the entity name, the nature and percentage of Your interest, dates the interest was acquired and, if applicable, divested, value of the interest, and whether or when such interest was disclosed to and approved by the OpenAI Board and regulators.\n>\n> **SUPPLEMENTAL RESPONSE TO INTERROGATORY NO. 4:**\n>\n> Subject to and without waiving the foregoing objections, in response to Interrogatory No. 4, pursuant to Rule 33(d) of the Federal Rules of Civil Procedure, Brockman states that he holds, or has held, direct or indirect financial interests in the following entities with which OpenAI, Inc. or its affiliates have entered business transactions: **[redacted] Cerebras Systems, Inc. (\"Cerebras\"), [redacted] CoreWeave, Inc. (\"CoreWeave\"), [redacted] Helion Energy, Inc. (\"Helion\"), [redacted] and Stripe, Inc. (\"Stripe\").** The details of Brockman's direct or indirect financial interests in these entities are set forth in the following table:\n>\n> | Entity | Date Brockman First Acquired Interest | Brockman's Holding (as of date of valuation) | FMV of Brockman's Interest | Date of Valuation |\n> |---|---|---|---|---|\n> | [redacted] | [redacted] | [redacted] | [redacted] | [redacted] |\n> | **Cerebras** | **03/17/2017** | **77,727** | **$2,816,235.51** | 12/31/2025 |\n> | [redacted] | [redacted] | [redacted] | [redacted] | [redacted] |\n> | **CoreWeave**³ | **06/17/2019** | **6,995** | **$817,365.75** | 04/21/2026 |\n> | [redacted] | [redacted] | [redacted] | [redacted] | [redacted] |\n> | **Helion Energy** | **08/23/2023** | **5,978** | **$433,584.34** | 12/31/2025 |\n> | [redacted] | [redacted] | [redacted] | [redacted] | [redacted] |\n> | [redacted] | [redacted] | [redacted] | [redacted] | [redacted] |\n> | [redacted] | [redacted] | [redacted] | [redacted] | [redacted] |\n> | **Stripe** | **12/16/2010** | **7,480,264** | **$471,256,632** | 03/2026 |\n>\n> ³ Brockman acquired 52,660 Series B-1 Preferred shares of Weights and Biases, LLC on June 17, 2019. That interest converted to 6,995 shares of CoreWeave as a result of CoreWeave's acquisition of Weights and Biases in May 2025, in connection with which Brockman also received a cash distribution of $29,317.97.\n>\n> It is my understanding that OpenAI, Inc. or its affiliates have conducted the following business with the referenced entities:\n>\n> - **[redacted]**\n> - **In December 2025, OpenAI OpCo, LLC entered into a multi-year agreement with Cerebras. Pursuant to the agreement, OpenAI agreed to purchase 750 megawatts of AI inference compute capacity and related services, including collaboration on engineering development and integrations.**\n> - **[redacted]**\n> - **In March 2025, OpenAI OpCo, LLC entered into a Master Services Agreement with CoreWeave related to anticipated purchases of computing capacity. OpenAI made an order for computing capacity for a total estimated purchase price of $11.945 billion in March 2025, and this order was amended and restated in May 2025. OpenAI made another order for an estimated purchase price of $4.021 billion in May 2025. As part of this transaction, OpenAI OpCo, LLC was issued CoreWeave Class A common stock valued at $350,000,000 through a private placement. In addition, in March 2025, OpenAI OpCo, LLC entered into an intellectual property agreement with CoreWeave and an agreement with respect to financing with CoreWeave, Dell Financial Services, LLC, and MUFG Bank, Ltd. In May 2025, OpenAI and CoreWeave announced an expanded agreement worth up to about $4 billion, and in September 2025, OpenAI and CoreWeave announced an additional expanded agreement worth up to approximately $6.5 billion.**\n> - **[redacted block — three sub-bullets]**\n> - **In May 2024, OpenAI OpCo, LLC entered into a scaled deployment acceleration agreement with Helion. That agreement creates no binding commitments and instead establishes a roadmap for potential future agreements between Helion and OpenAI, including potential arrangements to scale the manufacturing and deployment of Helion generators for the purpose of generating power to be purchased by OpenAI.**\n> - **In March 2026, OpenAI OpCo, LLC entered into a restated scaled deployment acceleration agreement with Helion. That restated agreement extends the term of the parties' existing collaboration. It provides a framework for potential future arrangements relating to Helion fusion projects that could supply power for OpenAI's AI infrastructure.**\n> - **[redacted block — six sub-bullets]**\n> - **In June 2023, OpenAI, LLC entered into a services agreement with Stripe pursuant to which Stripe agreed to provide payment-related services. Under that agreement, OpenAI, LLC initially committed to spend a minimum of $2.5 million per year over a three-year period. The June 2023 services agreement, and its associated fee schedule, were amended several times. In March 2025, OpenAI, LLC and Stripe entered into an amended fee schedule, which involved a three-year commitment in which OpenAI, LLC would pay Stripe $15 million for the first year, $20 million for the second year, and $25 million for the third year.**\n>\n> Brockman has disclosed his interests in third-party entities in accordance with OpenAI, Inc.'s conflict of interest policy.\n>\n> By way of further response, pursuant to Rule 33(d) of the Federal Rules of Civil Procedure, Brockman refers to his document production, including in response to RFP Nos. 31 and 33, from which the answer to this Interrogatory may be derived or ascertained.\n>\n> Date: April 29, 2026 — *MORRISON & FOERSTER LLP* — **/s/ Jordan Eth**\n>\n> **VERIFICATION:** I, Gregory Brockman, hereby verify under penalty of perjury under the laws of the United States that the responses contained in the foregoing GREGORY BROCKMAN'S FOURTH AMENDED SUPPLEMENTAL RESPONSES AND OBJECTIONS TO PLAINTIFF MUSK'S FIRST SET OF INTERROGATORIES are true and correct to the best of my knowledge, information, and belief based on a reasonable investigation. I make no verification with respect to any objections stated, or incorporated by reference, within the Fourth Amended Supplemental Responses and Objections.\n>\n> Executed on this 29th day of April 2026.\n>\n> **By: /s/ Gregory Brockman** [DocuSigned, 7F90F2F12AA746E…]\n\n## Commentary\n\nPX 390 is the documentary spine of plaintiffs' [[Key Themes]] § \"Greg Brockman's contributions ('zero')\" Cerebras-conflict thread. The verified table — Cerebras stock acquired **March 17, 2017** for an FMV of **$2.8M** at year-end 2025 — corroborates Day 6 testimony that Brockman bought into Cerebras while OpenAI was actively considering a Cerebras *merger* in 2017 (see [[PX-157]] for Sutskever and Brockman's own contemporaneous note: \"*In the event we decide to buy Cerebras, my strong sense is that it'll be done through Tesla*\"). The same conflict carries through to OpenAI OpCo's **December 2025 agreement to purchase 750 megawatts of Cerebras inference compute** — disclosed here in Brockman's own verified words and consistent with the Day 6 testimony that Cerebras' valuation went from ~$8B to ~$23B after the deal. The CoreWeave block adds a second axis: a **$11.945B March 2025 master services order** and a **$350M private placement** of CoreWeave Class A common to OpenAI OpCo, paired with Brockman's own holding (acquired indirectly via the Weights & Biases acquisition). Helion (whose CEO is Sam Altman's longtime portfolio company) and Stripe (Brockman's $471M holding from December 2010) round out the picture: every major OpenAI commercial counterparty with public visibility in 2025–2026 is a Brockman portfolio company. Brockman's verification — \"in accordance with OpenAI, Inc.'s conflict of interest policy\" — is the line plaintiffs will press into the disclosure-failure narrative on Day 7 redirect (the Cerebras 2017 disclosure failure, the March 2017 timing, etc.). The redacted entity rows (four to six additional companies still under seal) suggest the public picture is a partial subset of Brockman's full conflict map.\n\n---\n*See also:* [[Day 6|Day 6 digest]] · [[Day 7|Day 7 digest]] · [[Greg Brockman]] · [[PX-157]] · [[PX-161]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-5", "exhibit": "PX 5", "party": "Plaintiffs", "type": "Email", "admitted_trial_day": "Day 2 (April 28, 2026)", "uploaded_box_pt": "2026-04-28T14:51:56", "uploader": "Someone", "pages": 1, "size_bytes": 148845, "source_pdf": "PX-5.pdf", "pdf_url": "https://media.mts-in.com/PX-5.pdf", "body_markdown": "# PX 5 — June 25, 2015 Musk-Altman \"AI lab\" email\n\n> Sam Altman's June 24, 2015 founding pitch to Elon Musk for what would become OpenAI, with Musk's two-word reply: \"Agree on all.\"\n\n## Document type\n**Email, plain text, single message with quoted reply.** A one-page, two-message exchange: Altman's longer June 24, 2015 founding-pitch email to Musk laying out the proposed AI lab's mission, governance, team size, and Musk's role; and Musk's same-thread response sent the next morning (\"Agree on all\"). Both addresses redacted other than Musk's `erm@spacex.com` and Altman's `shga@ycombinator.com`-era handle. Bates 2024MUSK-0008261.\n\n## Logistics\n- **Trial admission:** Day 2 (April 28, 2026), opening day of testimony — used during Mr. Musk's direct examination as one of the earliest documents in plaintiffs' founding-narrative chronology. No prior wiki reference; admission day inferred from Box upload date.\n- **Box upload:** 2026-04-28 14:51:56 PT — uploaded same day it was used at trial, in the Day 2 morning batch.\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~145 KB, 1 page.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `005.pdf`.\n\n## Transcribed text\n\n> **From:** \"Elon Musk\" <erm@spacex.com>\n> **To:** \"Sam Altman\" [redacted]\n> **Subject:** Re: AI lab\n> **Date:** Thu, 25 Jun 2015 06:05:43 -0000\n> **Importance:** Normal\n>\n> Agree on all\n>\n> ---\n>\n> On Jun 24, 2015, at 10:24 AM, Sam Altman [redacted] wrote:\n>\n> ) The mission would be to create the first general AI and use it for individual empowerment—ie, the distributed version of the future that seems the safest. More generally, safety should be a first-class requirement.\n>\n> ) I think we'd ideally start with a group of 7-10 people, and plan to expand from there. We have a nice extra building in Mountain View they can have.\n>\n> ) I think for a governance structure, we should start with 5 people and I'd propose you, Bill Gates, Pierre Omidyar, Dustin Moskovitz, and me. The technology would be owned by the foundation and used \"for the good of the world\", and in cases where it's not obvious how that should be applied the 5 of us would decide. The researchers would have significant financial upside but it would be uncorrelated to what they build, which should eliminate some of the conflict (we'll pay them a competitive salary and give them YC equity for the upside). We'd have an ongoing conversation about what work should be open-sourced and what shouldn't. At some point we'd get someone to run the team, but he/she probably shouldn't be on the governance board.\n>\n> ) Will you be involved somehow in addition to just governance? I think that would be really helpful for getting work pointed in the right direction getting the best people to be part of it. Ideally you'd come by and talk to them about progress once a month or whatever. We generically call people involved in some limited way in YC \"part-time partners\" (we do that with Peter Thiel for example, though at this point he's very involved) but we could call it whatever you want. Even if you can't really spend time on it but can be publicly supportive, that would still probably be really helpful for recruiting.\n>\n> ) I think the right plan with the regulation letter is to wait for this to get going and then I can just release it with a message like \"now that we are doing this, I've been thinking a lot about what sort of constraints the world needs for safety.\" I'm happy to leave you off as a signatory. I also suspect that after it's out more people will be willing to get behind it.\n>\n> am\n\n## Commentary\n\nPX 5 is the founding handshake — Altman's June 24, 2015 pitch laying out the four-part architecture plaintiffs say became OpenAI: nonprofit foundation ownership, \"for the good of the world\" mission, safety as a first-class requirement, and Musk in a public-facing supporter/governance role. Musk's two-word \"Agree on all\" reply is exactly the assent plaintiffs build their charitable-trust theory on — it predates the December 2015 incorporation by six months and, on plaintiffs' theory, is the moment the parties met minds on terms that the 2018 LP conversion ([[DX 827]]) and the 2019 Microsoft license ([[PX 1504]]) later dismantled. The defense's reading on cross would emphasize that this is a high-level pitch deck in prose, not a contract, and contains no restriction on a future for-profit subsidiary. The \"Bill Gates, Pierre Omidyar, Dustin Moskovitz\" governance roster never materialized as written.\n\n---\n*See also:* [[Day 2|Day 2 digest]] · [[Elon Musk]] · [[Sam Altman]] · [[Key Themes]] · [[OpenAI Charter|PX 24]]\n"}
{"exhibit_id": "PX-51", "exhibit": "PX 51", "party": "Plaintiffs", "type": "IRS Form 990", "admitted_trial_day": "Day 4 (April 30, 2026)", "uploaded_box_pt": "2026-04-30T15:27:20", "uploader": "Someone", "pages": 43, "size_bytes": 5679199, "source_pdf": "PX-51.pdf", "pdf_url": "https://media.mts-in.com/PX-51.pdf", "body_markdown": "# PX 51 — OpenAI Inc. 2016 Form 990\n\n> OpenAI Inc.'s first full-year IRS Form 990 (calendar year 2016) — the foundational public-facing tax document representing OpenAI as a 501(c)(3) public charity, listing Musk and Altman as directors with $0 compensation, Brockman as CTO at $175K, and Ilya Sutskever as Research Director at $1.9M base + $1M bonus.\n\n## Document type\n**IRS Form 990 (Return of Organization Exempt From Income Tax)** for tax year 2016, with Schedules A, B, D, J, L, M, O, R, and Form 8868 extension. Filed for OpenAI Inc., EIN 81-0861541, 3180 18th St., Suite 100, San Francisco, CA 94110. Signed by Chris Clark, Dir/COO. Bates OPENAI_MUSK00009189–9231.\n\n## Logistics\n- **Trial admission:** Day 4 (April 30, 2026) — used during Musk cross + redirect and Birchall direct + cross to anchor OpenAI's public representations as a \"public charity… not a private foundation\" during the period Musk was donating.\n- **Box upload:** 2026-04-30 15:27:20 PT.\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~5.4 MB, 43 pages (form + schedules).\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `051.pdf`.\n\n## Transcribed text\n\n> **Form 990 — Return of Organization Exempt From Income Tax** (2016)\n>\n> **A.** For the 2016 calendar year. **B.** Address change. **C.** OpenAI Inc., 3180 18th St., Suite 100, San Francisco, CA 94110. **D. EIN:** 81-0861541. **E.** 833-927-2677. **F.** Principal officer: Chris Clark. **G. Gross receipts:** $13,807,074. **I. Tax-exempt status:** 501(c)(3). **J. Website:** openai.com. **K.** Corporation. **L. Year of formation:** 2015. **M. State of legal domicile:** DE.\n>\n> **Part I — Summary**\n> - Voting members of governing body: 5 (independent: 4)\n> - Total individuals employed CY2016: 52\n> - **Contributions and grants:** $13,784,637\n> - **Investment income:** $44\n> - **Other revenue:** $22,393\n> - **Total revenue:** $13,807,074\n> - Salaries / compensation / employee benefits: $7,056,443\n> - Other expenses: $4,181,233\n> - **Total expenses:** $11,237,676\n> - **Revenue less expenses:** $2,569,398\n> - End-of-year total assets: $2,662,055\n> - Total liabilities: $92,657\n> - **Net assets / fund balances:** $2,569,398\n>\n> Signed by **Chris Clark, Dir/COO**. Preparer: Michael Fontanello, Fontanello, Duffield & Otake, LLP, San Francisco.\n>\n> **Schedule O — Mission Statement**\n>\n> > OpenAI's goal is to advance digital intelligence in the way that is most likely to benefit humanity as a whole, unconstrained by a need to generate financial return. We think that artificial intelligence technology will help shape the 21st century, and we want to help the world build safe AI technology and ensure that AI's benefits are as widely and evenly distributed as possible. We're trying to build AI as part of a larger community, and we want to openly share our plans and capabilities along the way.\n>\n> **Part III — Program Service Accomplishments (Line 4a):**\n>\n> > In 2016, OpenAI established its research team, set initial goals, and chose its first major research projects. Accomplishments include launching the OpenAI Gym Beta, publishing nearly half a dozen comprehensive research papers, holding a self-organized machine learning conference, developing infrastructure for deep learning, and building a safety team.\n>\n> Program service expenses: $10,554,876.\n>\n> **Part VII — Compensation of Officers, Directors, Trustees, Key Employees**\n>\n> | Name | Title | Hours/wk | Position | W-2/1099 Org | W-2/1099 Related | Other |\n> |---|---|---|---|---|---|---|\n> | Elon Musk | Director | 3 | X | $0 | $0 | $0 |\n> | Sam Altman | Director | 5 | X | $0 | $0 | $0 |\n> | Chris Clark | Dir/COO | 20 | X (officer) | $40,942 | $164,871 | $4,642 |\n> | Jonathan Levy | Sec/Treasurer | 2 | X | $0 | $0 | $0 |\n> | Gregory Brockman | CTO | 40 | (officer) | $175,000 | $0 | $5,801 |\n> | Ilya Sutskever | Research Director | 40 | (key employee) | $1,900,000 | $0 | $12,282 |\n> | Ian Goodfellow | Research Scientist | 40 | (highest-comp.) | $808,243 | $0 | $11,619 |\n> | Pieter Abbeel | Technical Staff | 40 | (highest-comp.) | $425,000 | $0 | $4,360 |\n> | Man Wai Vicki Cheung | Research Engineer | 40 | (highest-comp.) | $297,917 | $0 | $5,851 |\n> | John Schulman | Senior Researcher | — | (highest-comp.) | $275,000 | $0 | $5,596 |\n> | Diederik Kingma | Technical Staff | 40 | (highest-comp.) | $172,917 | $0 | $9,516 |\n>\n> **Schedule J — Compensation breakdown (Sutskever):** Base $900,000 + bonus $1,000,000 + nontaxable benefits $12,282 = $1,912,282. (Goodfellow: base $208,243 + bonus $600,000.)\n>\n> **Schedule B — Schedule of Contributors**\n>\n> | # | Name | Total contributions | Type |\n> |---|---|---|---|\n> | 1 | Elon Musk granted via YC ORG, 335 Pioneer Way, Mountain View, CA 94041 | $10,000,000 | Person |\n> | 2 | Sam Altman, 3180 18th St., Suite 100, San Francisco, CA 94110 | $3,784,637 | Noncash (Forgiveness of Debt, FMV $3,784,637 dated 12/31/16) |\n>\n> **Schedule L Part II — Loans to/from Interested Persons**\n> - Sam Altman, Board Member/Officer — Loan from organization, Operations purpose, original principal $3,750,000.\n>\n> **Schedule R — Related Tax-Exempt Organizations**\n> - Y Combinator Research Inc., 469 9th Street, 2nd Floor, Oakland, CA 94607, EIN 81-0861414, 501(c)(3) public charity.\n>\n> **Part VI — Governance:** Conflict-of-interest policy: Yes. Whistleblower policy: No. Document retention/destruction policy: No.\n>\n> **Public charity status (Schedule A):** Box 7 checked — section 170(b)(1)(A)(vi), publicly supported organization. First five years box checked.\n\n## Commentary\n\nPX 51 is the **first complete public picture** OpenAI Inc. presented to the IRS — and to potential donors — about what it actually was in 2016. Plaintiffs use it for two purposes: (i) to anchor OpenAI's representation that it was a charitable, publicly supported 501(c)(3) (\"unconstrained by a need to generate financial return\"), which underpins their charitable-trust theory (see [[Key Themes]]); and (ii) to highlight Sam Altman's $3.78M \"Forgiveness of Debt\" noncash contribution and the contemporaneous $3.75M loan-from-OpenAI to Altman recorded in Schedule L — a related-party transaction plaintiffs flagged on Day 4 of trial. Note also that **Brockman is on the books as CTO at $175K**, which contrasts with the [[PX-8]] Oct 2015 representation that he would \"personally\" donate $100K, and stands alongside [[Brockman Journal]] entries from 2017 valuing his upside at \"$1,000,000,000.\" The 990 lists Sutskever's bonus at $1,000,000 — directly relevant to the \"hidden equity\" texture of [[Key Themes]] §\"Greg Brockman's contributions ('zero').\" The form was prepared in cash-vs-accrual accrual basis and reviewed by the COO before board finance-committee approval.\n\n---\n*See also:* [[Day 4|Day 4 digest]] · [[PX-112]] · [[Sam Altman]] · [[Chris Clark]] · [[Greg Brockman]] · [[Ilya Sutskever]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-52", "exhibit": "PX 52", "party": "Plaintiffs", "type": "IRS Form 990", "admitted_trial_day": "Day 4 (April 30, 2026)", "uploaded_box_pt": "2026-04-30T15:27:21", "uploader": "Someone", "pages": 44, "size_bytes": 6190387, "source_pdf": "PX-52.pdf", "pdf_url": "https://media.mts-in.com/PX-52.pdf", "body_markdown": "# PX 52 — OpenAI Inc. 2017 IRS Form 990\n\n> OpenAI Inc.'s 2017 federal Form 990 — the foundational nonprofit-tax-return exhibit showing $33.2M in 2017 contributions, the eight-director board (Musk, Altman, Brockman, Sutskever, Karnofsky, Clark, Levy, Lansky), the $3M Sam Altman officer loan, the Schedule B contributor list (Musk-side $18.1M via Foundation/Fidelity/Vanguard plus the $5M Aphorism and $10M Good Ventures gifts), and Brockman's $271K reported compensation.\n\n## Document type\n**IRS Form 990, calendar year 2017, full return with all schedules.** 44 pages including Schedules A, B, D, J, L, M, O, R and Form 8868 extension. Filed by Chris Clark (Dir/COO/Treas), prepared by Fontanello, Duffield & Otake LLP. Bates OPENAI_MUSK00006040–6083; stamped \"Confidential.\"\n\n## Logistics\n- **Trial admission:** Day 4 (April 30, 2026). Used during Birchall direct on the donation-by-donation walk-through and the public-charity contributor structure (see [[PX-87]] tax acknowledgment for the May 2017 $5M wire and [[Key Themes]] § \"Greg Brockman's contributions ('zero')\").\n- **Box upload:** 2026-04-30 15:27:21 PT — late-afternoon Day 4 batch (clustered with PX 75, PX 87, PX 92).\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~5.9 MB, 44 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `052.pdf`.\n\n## Transcribed text\n\n> **Form 990 — Return of Organization Exempt From Income Tax — 2017**\n> **Organization:** OpenAI Inc. — 3180 18th St., Suite 100, San Francisco, CA 94110\n> **EIN:** 81-0861541 · **Phone:** 833-927-2677 · **Website:** openai.com\n> **State of legal domicile:** DE · **Year of formation:** 2015 · **501(c)(3)**\n> **Gross receipts:** $33,228,665\n>\n> **Part I — Summary**\n> - Voting members of governing body (line 1a): **8**\n> - Independent voting members (1b): **2**\n> - Total individuals employed in 2017 (line 5): **99**\n> - Contributions and grants (line 8): prior year $13,784,637 → current year $33,228,555\n> - Total revenue (line 12): $33,228,665\n> - Salaries, other compensation, employee benefits (line 15): $15,775,142\n> - Total expenses (line 18): $28,668,228\n> - Revenue less expenses (line 19): $4,560,437\n> - End-of-year total assets (line 20): $10,488,224\n> - End-of-year net assets (line 22): $7,129,835\n>\n> **Part III — Statement of Program Service Accomplishments**\n> \"In 2017 OpenAI demonstrated that reinforcement learning algorithms could be scaled to beat the world's best humans at a restricted version of an advanced, multiplayer game called Dota2. The organization also took part in a multi-stakeholder report on the potential malicious uses of AI, and published its findings.\"\n> Total program service expenses: $28,459,690.\n>\n> **Part VII — Officers, Directors, Trustees, Key Employees, and Highest Compensated Employees**\n>\n> | Name | Title | Hrs/wk | Reportable comp (W-2) | Related-org comp | Other comp |\n> |---|---|---|---|---|---|\n> | Elon Musk | Director | 3 | $0 | $0 | $0 |\n> | Sam Altman | Director | 5 | $0 | $0 | $0 |\n> | Chris Clark | Dir/COO/Treas | 20 | $187,500 | $150,000 | $7,424 |\n> | Holden Karnofsky | Director | 5 | $0 | $0 | $0 |\n> | Gregory Brockman | Director/CTO | 40 | $264,201 | $0 | $6,801 |\n> | Ilya Sutskever | Research Dir. | 40 | $748,908 | $0 | $14,521 |\n> | Jonathan Levy | Secretary | 2 | $0 | $0 | $0 |\n> | David Lansky | General Counsel | 20 | $173,397 | $52,019 | $17,108 |\n> | Andrej Karpathy | Technical Staff | 40 | $424,231 | $0 | $3,355 |\n> | Pieter Abbeel | Technical Staff | 40 | $658,077 | $0 | $4,605 |\n> | Wojciech Zaremba | Research Engineer | 40 | $508,584 | $0 | $7,009 |\n> | John Schulman | Senior Researcher | 0 | $718,728 | $0 | $6,577 |\n> | Diederik Kingma | Technical Staff | 40 | $545,833 | $0 | $12,509 |\n>\n> Total individuals receiving more than $100,000 of reportable compensation: **50.**\n>\n> **Schedule B — Schedule of Contributors (2017)**\n>\n> | # | Donor | Total | Type |\n> |---|---|---|---|\n> | 1 | Aphorism Foundation (314 Lytton Ave #200, Palo Alto) | $5,000,000 | Person |\n> | 2 | Elon Musk granted via Fidelity (200 Seaport Blvd, Boston) | $1,140,000 | Person |\n> | 3 | Good Ventures Foundation (2440 W El Camino Real, Mountain View) | $10,000,000 | Person |\n> | 4 | Elon Musk Via Vanguard Charitable (Warwick, RI) | $700,000 | Person |\n> | 5 | Musk Foundation via YC ORG (335 Pioneer Way, Mountain View) | $16,028,500 | Person |\n> | 6 | Elon Musk (PO Box 10195 Dept 863, Palo Alto) | $248,295 | Noncash — 4 Vehicles (FMV) |\n> | 7 | Donor's Trust DAF (Alexandria, VA) | $100,000 | Person |\n>\n> **Schedule L — Loans to and/or From Interested Persons**\n> Sam Altman, Board Member/Officer · Purpose: Operations · Loan **from** Sam Altman to OpenAI · Original principal: **$3,000,000** · Balance due: **$3,006,575** · Approved by board, written agreement.\n>\n> **Schedule R — Related Organizations**\n> Y Combinator Research Inc. (469 9th St, 2nd Fl, Oakland, CA 94607 · EIN 81-0861414) · 501(c)(3) · charitable.\n>\n> **Mission (Schedule O):**\n> \"OpenAI's goal is to advance digital intelligence in the way that is most likely to benefit humanity as a whole, unconstrained by a need to generate financial return. We think that artificial intelligence technology will help shape the 21st century, and we want to help the world build safe AI technology and ensure that AI's benefits are as widely and evenly distributed as possible. We're trying to build AI as part of a larger community, and we want to openly share our plans and capabilities along the way.\"\n\n## Commentary\n\nPX 52 is the cleanest single document for plaintiffs' charitable-trust theory in 2017: a sworn IRS filing, signed by Chris Clark, declaring under penalty of perjury that OpenAI is a 501(c)(3) \"unconstrained by a need to generate financial return,\" that its Schedule B contributors are personally and DAF-routed Musk gifts and a small handful of mission-aligned foundations (Aphorism, Good Ventures, Donor's Trust). The form also nails several discrete trial facts: that **Brockman's reportable comp in 2017 was $271K** (the \"discount\" salary referenced in [[PX-161]] — \"people join openai at a non-profit discount\"), that **Sutskever was paid $763K** (well above the \"non-profit discount\"), that the **$3M Altman loan** was on the books through year-end, and that the entity had **99 employees** with 50 over $100K — i.e., a real organization, not a shell. The defense reading: every Musk wire on the contributor list is a deductible charitable contribution to a sister-organization-supported 501(c)(3), supporting the [[Key Themes]] § \"donor-advised funds / tax-deduction wrinkle\" cross of Birchall. The contributor-table format (Musk routed via Fidelity, Vanguard, and the Musk Foundation through YC.org) is the documentary backbone of the **$38,191,066** total Birchall confirmed on Day 4 (PX 112A).\n\n---\n*See also:* [[Day 4|Day 4 digest]] · [[PX-87]] · [[PX-71]] · [[Jared Birchall]] · [[Sam Altman]] · [[Greg Brockman]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-53", "exhibit": "PX 53", "party": "Plaintiffs", "type": "IRS Form 990", "admitted_trial_day": "unknown", "pages": 84, "size_bytes": 6156412, "source_pdf": "PX-53.pdf", "pdf_url": "https://media.mts-in.com/PX-53.pdf", "body_markdown": "# PX 53 — OpenAI Inc. 2018 Form 990\n\n> OpenAI Inc.'s 2018 IRS Form 990 — the tax-year covering the months immediately preceding the **March 2019 OpenAI LP launch**, when contributions and grants spiked to **$49.9M** and the nonprofit's total expenses crossed **$51.5M** for the first time.\n\n## Document type\n**IRS Form 990 (Return of Organization Exempt From Income Tax)** for tax year 2018. Filed for OpenAI Inc., EIN **81-0861541**, 3180 18th St., Suite 100, San Francisco, CA 94110. Phone (415) 879-9686.\n\n## Logistics\n- **Trial admission:** unknown (not yet matched to a specific day in the public record).\n- **File size:** ~5.9 MB, 84 pages (form + schedules).\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `053.pdf`.\n\n## Transcribed text\n\n> **Form 990 — Return of Organization Exempt From Income Tax** (2018 calendar year)\n>\n> **Part I — Summary**\n> - Voting members of governing body: **8** (independent: 6)\n> - Total individuals employed CY2018: **113**\n> - **Contributions and grants:** $49,917,797\n> - **Investment income:** $50,834\n> - **Total revenue:** $49,968,631\n> - Salaries / compensation / employee benefits: $15,406,921\n> - Other expenses: $36,142,088\n> - **Total expenses:** $51,549,009\n> - **Revenue less expenses:** $(1,580,378)\n> - **Total assets (EOY):** $25,926,281; total liabilities $23,753,896.\n>\n> **Schedule B — Schedule of Contributors** (selected, 2018):\n> | # | Name | Total contributions | Type |\n> |---|---|---|---|\n> | 1 | **Aphorism Foundation**, 314 Lytton Ave #200, Palo Alto, CA 94301 | $5,000,000 | Person |\n> | 2 | **Fidelity Charitable**, 200 Seaport Blvd NCW4B, Boston, MA 02210 | $3,630,000 | Person |\n>\n> **Noncash contributions** included in lines 1a–1f: **$5,619,531**.\n>\n> [Remainder: Schedule O mission narrative, Schedule J officer-comp breakdown, Schedule L related-party transactions, Schedule R related orgs — pages 12–84 not transcribed here.]\n\n## Commentary\n\nPX 53 picks up where [[PX 51]] (2016) and [[PX 52]] (2017) left off. The 2018 990 is the **last full year before** OpenAI Inc. converted to the LP-and-nonprofit-parent structure announced in March 2019 (see [[PX 241]]) — the year **headcount jumps to 113** and the nonprofit's expenses cross **$51M**. The 2018 contributors list shifts away from direct Musk Foundation grants (compare [[PX 60]]) toward institutional DAFs and a new entrant — the **Aphorism Foundation** ($5M) — which plaintiffs' theory uses to color the period as one in which Musk-vehicle giving is winding down even as expenses spike, foreshadowing the for-profit pivot.\n\n---\n*See also:* [[PX 51]] · [[PX 52]] · [[PX 54]] · [[PX 241]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-54", "exhibit": "PX 54", "party": "Plaintiffs", "type": "IRS Form 990", "admitted_trial_day": "unknown", "pages": 88, "size_bytes": 6183311, "source_pdf": "PX-54.pdf", "pdf_url": "https://media.mts-in.com/PX-54.pdf", "body_markdown": "# PX 54 — OpenAI Inc. 2019 Form 990\n\n> OpenAI Inc.'s 2019 IRS Form 990 — the tax-year of the **OpenAI LP launch and the $1B Microsoft investment**. Salaries collapse from **$15.4M to $616K** as research staff move to OpenAI LP, but the 990 still records **$33.6M in contributions** and reports **$1M in real-estate rent income**.\n\n## Document type\n**IRS Form 990** for tax year 2019. OpenAI Inc., EIN 81-0861541, San Francisco, CA.\n\n## Logistics\n- **Trial admission:** unknown.\n- **File size:** ~5.9 MB, 88 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `054.pdf`.\n\n## Transcribed text\n\n> **Part I — Summary** (2019 calendar year)\n> - Voting members of governing body: **8** (independent: 6)\n> - Total individuals employed CY2019: **125**\n> - **Contributions and grants:** $33,580,000\n> - **Other revenue:** $(1,647,271)\n> - **Total revenue:** $31,932,729\n> - Grants and similar amounts paid: $150,000\n> - Salaries / compensation / employee benefits: **$615,645** (vs. $15.4M in 2018)\n> - Other expenses: $2,567,272\n> - **Total expenses:** $3,332,917\n> - **Revenue less expenses:** $28,599,812\n> - **Total assets (EOY):** $31,040,138.\n> - **Real estate rents (Part VIII):** $1,000,000.\n>\n> **Schedule B — Contributors** (selected, 2019):\n> | # | Name | Total contributions |\n> |---|---|---|\n> | 1 | **Fidelity Charitable**, P.O. Box 770001, Cincinnati, OH 45277 | $3,480,000 |\n> | 2 | **Amazon Web Services**, 410 Terry Avenue N | $100,000 |\n>\n> [Bulk: Schedule O 2019 mission/program-service narrative; Schedule J officer comp; Schedule R related orgs (now including OpenAI LP / OpenAI GP, LLC). Pages 12–88 not transcribed here.]\n\n## Commentary\n\nPX 54 captures the **2019 inflection year** in 990 form. The ~25× drop in salaries (from $15.4M to $616K) is the paper-trail signature of OpenAI Inc. transferring its operating research team and operations to the new **OpenAI L.P.** in mid-2019 — see [[PX 200]] (October 2018 LP Agreement), [[PX 201]] (July 2019 amended LP Agreement), [[PX 202]] (July 2019 JDCA). Contributions still flow into the nonprofit at $33.6M in 2019, and the **$1M real-estate rent income line** is a notable artifact of the Pioneer Building economics surfaced in [[PX 79]] / [[DX 539]] / [[DX 600]]. **Headcount on the books still reads 125** even as the salaries number collapses — a structural curiosity that may reflect timing of the transition.\n\n---\n*See also:* [[PX 53]] · [[PX 55]] · [[PX 200]] · [[PX 201]] · [[PX 202]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-55", "exhibit": "PX 55", "party": "Plaintiffs", "type": "IRS Form 990", "admitted_trial_day": "unknown", "pages": 90, "size_bytes": 6300371, "source_pdf": "PX-55.pdf", "pdf_url": "https://media.mts-in.com/PX-55.pdf", "body_markdown": "# PX 55 — OpenAI Inc. 2020 Form 990\n\n> OpenAI Inc.'s 2020 IRS Form 990 — the first full year as a **nonprofit parent** of OpenAI LP. Contributions drop to **$2.66M** and the nonprofit pays out **$10.25M in grants** — its largest grant-making year before or since.\n\n## Document type\n**IRS Form 990** for tax year 2020. OpenAI Inc., EIN 81-0861541, San Francisco, CA.\n\n## Logistics\n- **Trial admission:** unknown.\n- **File size:** ~6.0 MB, 90 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `055.pdf`.\n\n## Transcribed text\n\n> **Part I — Summary** (2020 calendar year)\n> - Voting members of governing body: **8** (independent: 4)\n> - Total individuals employed CY2020: **11** (post-LP transition)\n> - **Contributions and grants:** $2,661,461\n> - **Investment income:** $305,323\n> - **Other revenue:** $515,000\n> - **Total revenue:** $3,481,784\n> - **Grants and similar amounts paid:** $10,250,005\n> - Salaries / compensation / employee benefits: $881,719\n> - Other expenses: $1,858,819\n> - **Total expenses:** $12,990,543\n> - **Revenue less expenses:** $(9,508,759)\n> - **Total assets (EOY):** $21,376,567; total liabilities $113,129.\n>\n> **Schedule B — Contributors** (selected, 2020):\n> | # | Name | Total contributions |\n> |---|---|---|\n> | 1 | **Fidelity Charitable**, 200 Seaport Blvd NCW4B, Boston, MA | $2,610,000 |\n>\n> [Pages 12–90: Schedule O mission narrative; Schedule J officer compensation; Schedule R related orgs (OpenAI LP, OpenAI GP, LLC, OpenAI OpCo); Form 8868 extension. Not transcribed here.]\n\n## Commentary\n\nPX 55 is the **first 990 of OpenAI Inc. as a stripped-down nonprofit parent**. Headcount drops to **11** (a number plaintiffs invoke when arguing the nonprofit is a thin shell). The **$10.25M grant outflow** is the largest single year of grant-making in the OpenAI Inc. 990 series and corresponds to a deliberate spend-down of pre-LP contribution accruals.\n\n---\n*See also:* [[PX 54]] · [[PX 56]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-56", "exhibit": "PX 56", "party": "Plaintiffs", "type": "IRS Form 990", "admitted_trial_day": "unknown", "pages": 76, "size_bytes": 5655812, "source_pdf": "PX-56.pdf", "pdf_url": "https://media.mts-in.com/PX-56.pdf", "body_markdown": "# PX 56 — OpenAI Inc. 2021 Form 990\n\n> OpenAI Inc.'s 2021 IRS Form 990 — contributions collapse to **$3,066** as the for-profit subsidiary takes over fundraising; total revenue is **~$23,456** against **$1.39M in expenses**.\n\n## Document type\n**IRS Form 990** for tax year 2021. OpenAI Inc., EIN 81-0861541, San Francisco, CA.\n\n## Logistics\n- **Trial admission:** unknown.\n- **File size:** ~5.4 MB, 76 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `056.pdf`.\n\n## Transcribed text\n\n> **Part I — Summary** (2021 calendar year)\n> - Voting members of governing body: **8** (independent: 4)\n> - Total individuals employed CY2021: **6**\n> - **Contributions and grants:** $3,066\n> - **Investment income:** $8,662\n> - **Other revenue:** $11,728\n> - Grants and similar amounts paid: $75,000\n> - Salaries / compensation / employee benefits: $327,614\n> - Other expenses: $986,950\n> - **Total expenses:** $1,389,564\n> - **Revenue less expenses:** $(1,377,836)\n> - **Total assets (EOY):** $19,976,363.\n>\n> [Pages 12–76: Schedule O mission narrative; Schedule J officer compensation; Schedule R related orgs. Not transcribed here.]\n\n## Commentary\n\nPX 56 is the **steady-state 990** that follows the 2020 grant-paydown year. Headcount on the books drops to **6**; contributions are functionally zero (**$3,066 total**). The contrast with the same year's enormous OpenAI LP fundraising activity is itself the point — see plaintiffs' nonprofit-shell framing in [[Key Themes]].\n\n---\n*See also:* [[PX 55]] · [[PX 57]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-57", "exhibit": "PX 57", "party": "Plaintiffs", "type": "IRS Form 990", "admitted_trial_day": "unknown", "pages": 88, "size_bytes": 6799023, "source_pdf": "PX-57.pdf", "pdf_url": "https://media.mts-in.com/PX-57.pdf", "body_markdown": "# PX 57 — OpenAI Inc. 2022 Form 990\n\n> OpenAI Inc.'s 2022 IRS Form 990 — the year of **ChatGPT's launch**. The nonprofit parent itself reports just **$44,485 in revenue** and **$1.29M in expenses**, including **$363K in grants paid**.\n\n## Document type\n**IRS Form 990** for tax year 2022. OpenAI Inc., EIN 81-0861541, San Francisco, CA.\n\n## Logistics\n- **Trial admission:** unknown.\n- **File size:** ~6.5 MB, 88 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `057.pdf`.\n\n## Transcribed text\n\n> **Part I — Summary** (2022 calendar year)\n> - Voting members of governing body: **9** (independent: 6)\n> - Total individuals employed CY2022: **2**\n> - **Contributions and grants:** $1,611\n> - **Investment income:** $42,874\n> - **Total revenue:** $44,485\n> - Grants and similar amounts paid: $363,333\n> - Salaries / compensation / employee benefits: $129,205\n> - Other expenses: $801,611\n> - **Total expenses:** $1,294,149\n> - **Revenue less expenses:** $(1,249,664)\n>\n> [Pages 12–88: Schedule O mission narrative; Schedule J officer compensation; Schedule R related orgs (now including OpenAI Global, LLC, OAI Corporation, LLC, etc.). Not transcribed here.]\n\n## Commentary\n\nPX 57 is the **ChatGPT-year 990 of the parent nonprofit**. The contrast is the story plaintiffs tell: the world-historical product is being shipped by OpenAI's for-profit subsidiaries, but the 501(c)(3) parent itself records **$1,611 in contributions** and just **2 employees on the books**. The November 2023 board firing is still in the future (see [[PX 304]]), but the structural concentration of activity inside the for-profit is already complete.\n\n---\n*See also:* [[PX 56]] · [[PX 58]] · [[PX 304]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-58", "exhibit": "PX 58", "party": "Plaintiffs", "type": "IRS Form 990", "admitted_trial_day": "unknown", "pages": 104, "size_bytes": 7324873, "source_pdf": "PX-58.pdf", "pdf_url": "https://media.mts-in.com/PX-58.pdf", "body_markdown": "# PX 58 — OpenAI Inc. 2023 Form 990\n\n> OpenAI Inc.'s 2023 IRS Form 990 — the year of the **November 2023 board firing**. **Schwab Fund for Charitable Giving** appears as a $5M contributor; the 990 records the **November 19/29, 2023 board departures of Greg Brockman and Ilya Sutskever**.\n\n## Document type\n**IRS Form 990** for tax year 2023. OpenAI Inc., EIN 81-0861541, San Francisco, CA.\n\n## Logistics\n- **Trial admission:** unknown.\n- **File size:** ~7.0 MB, 104 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `058.pdf`.\n\n## Transcribed text\n\n> **Part I — Summary** (2023 calendar year)\n> - Voting members of governing body: **3** (independent: 3)\n> - Total individuals employed CY2023: **2**\n> - **Contributions and grants:** $5,023,159\n> - **Investment income:** $340,089\n> - **Total revenue:** $5,363,248\n> - Grants and similar amounts paid: $2,641,712\n> - Other expenses: $225,134\n> - **Total expenses:** $2,866,846\n> - **Revenue less expenses:** $2,496,402\n> - **Total assets (EOY):** $22,678,253; total liabilities $1,545,913.\n>\n> **Schedule B — Contributors** (selected, 2023):\n> | # | Name | Total contributions |\n> |---|---|---|\n> | 1 | **Schwab Fund for Charitable Giving**, 211 Main Street, San Francisco, CA 94105 | $5,000,000 |\n>\n> **Officers / Directors (Part VII, 2023):**\n> - **Ilya Sutskever** — Director Through November 29, 2023.\n> - **Greg Brockman** — Director Through November 19, 2023.\n> - **Sam Altman** — Director & CEO Through November 19, 2023; CEO From November 29, 2023.\n>\n> [Pages 12–104: Schedule O mission narrative; Schedule J officer-compensation breakdown; Schedule L related-party transactions; Schedule R related orgs. Not transcribed here.]\n\n## Commentary\n\nPX 58 documents the **November 2023 board crisis on the IRS form itself**: Brockman's board departure on **Nov 19** (the day of the firing-week aftermath); Sutskever's departure on **Nov 29**; Altman recorded as CEO **Through Nov 19** and again **From Nov 29**, with a 10-day gap on the books that mirrors the public timeline. The 990's 2023 contributors line up with the period when, per the trial record, Altman's structural critique by the board ([[PX 304]]; [[Helen Toner]] / [[Tasha McCauley]] depositions on Days 8–9) was reaching its breaking point — see [[Day 8|Day 8 digest]] and [[Day 9|Day 9 digest]].\n\n---\n*See also:* [[PX 57]] · [[PX 304]] · [[PX 306]] · [[Day 8]] · [[Day 9]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-60", "exhibit": "PX 60", "party": "Plaintiffs", "type": "Compendium / multi-doc", "admitted_trial_day": "Day 4 (April 30, 2026)", "uploaded_box_pt": "2026-04-30T15:26:53", "uploader": "Someone", "pages": 69, "size_bytes": 2188254, "source_pdf": "PX-60.pdf", "pdf_url": "https://media.mts-in.com/PX-60.pdf", "body_markdown": "# PX 60 — Vanguard Charitable donor-advised-fund records: Musk Charitable Fund grants to OpenAI / YC ORG (2014–2017)\n\n> A compendium of Vanguard Charitable gift agreement, contribution confirmations, and grant confirmations documenting Musk's donor-advised-fund flow into OpenAI — the paper trail behind plaintiffs' $38M-contribution number and defense's \"donor-advised funds / tax-deduction\" theme.\n\n## Document type\n**Compendium / multi-doc** of Vanguard Charitable account documents: (1) the original 2014 \"Gift agreement for individuals or trusts\" establishing The Musk Charitable Fund (donor: Elon Musk Revocable Trust), (2) SolarCity and Tesla Motors stock contribution confirmations into the donor-advised fund, (3) grant confirmation letters from The Musk Charitable Fund / The Musk Foundation Charitable Fund to \"YC ORG\" (the original placeholder name of OpenAI Inc., 335 Pioneer Way, Mountain View) and to \"Openai Inc,\" and (4) a Vanguard Charitable Gemini case-management memo about diligence on the first $5M YC ORG grant. Confidential VC000240–VC000308.\n\n## Logistics\n- **Trial admission:** Day 4 (April 30, 2026) — Birchall direct + cross. Used to walk Birchall through the actual mechanics of how Musk's donations to OpenAI flowed (DAF → grant), reinforcing the precise contribution accounting Birchall confirmed at $38,191,066 in ~60 contributions (see [[Key Themes]] §\"$1B / $38M gap\").\n- **Box upload:** 2026-04-30 15:26:53 PT — Day 4 mid-afternoon Plaintiffs' batch (clustered with PX 79 and PX 93).\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~2.1 MB, 69 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `060.pdf`.\n\n## Transcribed text\n\nSelected substantive contents (the bulk of the 69-page PDF is repeating Vanguard Charitable boilerplate \"Important information about grants\" pages, summarized at the end).\n\n> **Vanguard Charitable Gift Agreement — for individuals or trusts** (faxed 07/10/2014)\n>\n> **1. Account Name** — The **Musk Charitable Fund**\n>\n> **2. Donor Information** — Trust: **Elon Musk Revocable Trust**, Trust Agreement Date 07-22-2003. Owner/Trustee A: Mr. Elon R. Musk, 2200 Geng Road, Suite 100, Palo Alto, CA 94303. Email: MUS001FOS@myCFO.com.\n\n> **Contribution confirmation — November 27, 2014** (Vanguard Charitable, signed Benjamin R. Pierce, President):\n> Contribution from **Elon Musk Revocable Trust** to support **The Musk Charitable Fund** — **363,000.0000 shares** of SolarCity Corporation, received 11/25/2014.\n\n> **Contribution confirmation — December 23, 2015** (signed Brad Caswell, COO):\n> **35,000.0000 shares** of SolarCity Corporation, received 12/21/2015, into The Musk Charitable Fund.\n\n> **Grant confirmation — June 3, 2016** (Janice L. Conner, CFO, Vanguard Charitable):\n> Grant from **The Musk Charitable Fund** to **YC ORG** (Mr. Chris Clark, 335 Pioneer Way, Mountain View, CA 94041), **$5,000,000.00**, To be used for: \"**The OpenAI Artificial Intelligence Research Program**.\" Recognized donors: Mr. Elon R. Musk; Mr. Ronald F. Gong.\n\n> **Vanguard Charitable Gemini case file — Case #C61878725** (\"Grant to YC Org\"; 05/25/2016 08:23):\n> \"**Brittany / Cindy — Elon Musk recommended a $5 million grant yesterday to a newly created organization called YC Org. It is in Guidestar but there is no website, email address or phone number. I want to call the organization to confirm that we have the correct mailing address and I also want to learn more about the purpose given the size of this grant (Purpose: 'The OpenAI Artificial Intelligence Research Program').** Is there any way that you can ask the donor to provide us with contact information for the organization?\"\n\n> **Contribution confirmations — Tesla Motors Inc** (into The Musk Foundation Charitable Fund, account #...537):\n> - 02/01/2017: 6,385.0000 shares ($1,593,833.83); 40,015.0000 shares ($9,985,018.96); 23,576.9320 shares ($5,887,442.65) — total ~$17.5M (Feb 8, 2017 confirmation).\n> - 02/01/2017: 20,023.0680 shares ($5,000,000.00) (Feb 6, 2017 confirmation).\n> - 11/08/2016: 26,000.0000 shares ($5,112,909.75) (Nov 16, 2016 confirmation).\n\n> **Grant confirmations — The Musk Foundation Charitable Fund → YC ORG / OpenAI Inc.** (the recurring pattern; Jane G. Greenfield, President):\n> - 08/19/2016 — **$4,500,000.00** to YC ORG, \"For 'The OpenAI Artificial Intelligence Research Program'.\"\n> - 09/22/2016 — **$142,000.00** to YC ORG, same purpose.\n> - 10/17/2016 — **$142,000.00** to YC ORG, same purpose.\n> - 11/11/2016 — **$750,000.00** to YC ORG, \"Open AI.\"\n> - 11/15/2016 — **$142,000.00** to YC ORG, same purpose.\n> - 11/30/2016 — **$4,250,000.00** to YC ORG, same purpose; \"By accepting this grant, you certify that your organization would not be classified a private non-operating foundation without the support it has received from Vanguard Charitable, including this grant.\"\n> - 12/15/2016 — **$142,000.00** to YC ORG.\n> - 01/17/2017 — **$142,000.00** to YC ORG.\n> - 02/15/2017 — **$142,000.00** to YC ORG.\n> - 03/16/2017 — **$175,000.00** to YC ORG, \"Department: **Openai March Lease Payment**.\"\n> - 04/18/2017 — **$175,000.00** to **Openai Inc**, \"Department: **Lease payment**.\"\n> - 05/15/2017 — **$175,000.00** to Openai Inc, \"Lease payment.\" cc'd to Mr. Jared J. Birchall, PO Box 49258, Los Angeles, CA 90049.\n> - 06/15/2017 — two **$175,000.00** grants to Openai Inc, \"Lease payment\" (separate confirmations to Mr. Elon R. Musk and Mr. Jared J. Birchall).\n\n> [Pages 5–6, 16–17, 20–21, 24–25, 28–29, 32–33, 36–37, 40–41, 44–45, 48–49, 52–53, 56–57, 60–61, 64–65, 68–69 — Vanguard Charitable's boilerplate \"Important information about grants\" pages: prohibited benefits, no bifurcation, pledges and other legal obligations, grant recognition, no additional charitable deduction, no lobbying, scholarships, private non-operating foundations, fiscal sponsors, disqualified supporting organizations, OFAC, additional information. Each grant letter is followed by 2 pages of these standard certifications, repeated verbatim ~15 times across the compendium.]\n\n## Commentary\n\nPX 60 is the documentary spine of plaintiffs' \"where did Musk's $38M actually come from and where did it go\" theory. Three structural facts emerge from the paper. **First**, Musk's contributions to OpenAI did not flow directly from Musk-the-individual; they flowed from **The Musk Charitable Fund** (later **The Musk Foundation Charitable Fund**), Vanguard Charitable donor-advised-fund accounts seeded with **SolarCity and Tesla stock** out of the Elon Musk Revocable Trust. This is the legal mechanic that makes the defense's \"tax-deduction\" theme bite — see [[Key Themes]] §\"donor-advised funds / tax-deduction wrinkle\" and Birchall's cross concession that \"for every single one of those donations, it's somewhere in the chain, Mr. Musk was entitled to take a tax deduction.\" **Second**, the recipient is repeatedly identified by the **placeholder name \"YC ORG\"** at 335 Pioneer Way, Mountain View, with **Chris Clark** as the contact — a useful corroboration of the early founding-period administrative entanglement with Y Combinator. **Third**, beginning in March 2017 the grant memos shift from \"The OpenAI Artificial Intelligence Research Program\" to \"**Lease payment**\" and \"**Openai March Lease Payment**\" — i.e., Musk's foundation is paying OpenAI's rent at the Pioneer Building, the same building leased by Musk Industries LLC under [[PX 79]] and discussed at length on Day 4 (see [[Key Themes]] §\"Pioneer Building\" and [[DX 539]], [[DX 600]]). The recurring **$175,000/month** grants in 2017 line up to the Pioneer Building base rent figure on Tenant's side of the master lease in PX 79. The Vanguard Charitable case-file note that YC ORG had \"no website, email address or phone number\" the day after Musk recommended a $5M grant to it is a reminder of how new the entity was when Musk's giving started.\n\n---\n*See also:* [[Day 4|Day 4 digest]] · [[PX 79]] · [[PX 103]] · [[DX 539]] · [[DX 600]] · [[Jared Birchall]] · [[Elon Musk]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-66", "exhibit": "PX 66", "party": "Plaintiffs", "type": "Email thread", "admitted_trial_day": "Day 6 (May 4, 2026)", "uploaded_box_pt": "2026-05-04T14:46:16", "uploader": "Someone", "pages": 6, "size_bytes": 3157507, "source_pdf": "PX-66.pdf", "pdf_url": "https://media.mts-in.com/PX-66.pdf", "body_markdown": "# PX 66 — Dec 2015 Brockman/Sutskever/Schulman recruit thread for Diederik Kingma\n\n> A six-page reply-chain email thread Greg Brockman forwarded to Sam Altman on December 6, 2015 — Diederik Kingma agreeing to join OpenAI (\"Elon's involvement is awesome. It's an honor to be part of this team\") on top of Schulman's and Brockman's earlier pitches to him, including a quoted private Musk email (\"I've decided to dedicate whatever amount of my time that is materially useful\").\n\n## Document type\n**Email thread, plain text, multi-message chain.** Brockman forwarding Ilya Sutskever's \"Strong, true words\" reaction on top of Kingma's \"I accept!\" message, on top of Brockman's, John Schulman's, and Kingma's prior back-and-forth from Nov 28 – Dec 6, 2015. Bates OPENAI_MUSK00016951–16956.\n\n## Logistics\n- **Trial admission:** Day 6 (May 4, 2026 — Russell direct + cross; Brockman cross by Mr. Kry). Per the Day 6 transcript table: \"Dec 2015 emails — Sutskever recruit Diederik Kingma joining OpenAI: 'Elon's involvement is awesome. It's an honor to be part of this team.'\" *(5/4/2026 Testimony @ ~09:55 PT)*\n- **Box upload:** 2026-05-04 14:46:16 PT — Day 6 mid-afternoon batch (clustered with PX 153, PX 164).\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~3.0 MB, 6 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `066.pdf`.\n\n## Transcribed text\n\n> **From:** Greg Brockman <gdb@openai.com>\n> **To:** Sam Altman <sam@ycombinator.com>\n> **Subject:** Fwd: Decision\n> **Date:** 12/6/2015 4:33:04 PM\n>\n> (Good timing on this + Sergey emails.)\n>\n> - gdb\n\n> **From:** Ilya Sutskever <ilyasu@openai.com>\n> **To:** Greg Brockman <gdb@openai.com>\n> **Subject:** Re: Decision\n> **Date:** Sun, Dec 6, 2015 at 7:26 AM\n>\n> Strong, true words.\n\n> **From:** Greg Brockman <gdb@openai.com>\n> **To:** Ilya Sutskever <ilyasu@openai.com>\n> **Subject:** Fwd: Decision\n> **Date:** Sun, Dec 6, 2015 at 10:16 AM\n>\n> [Greg forwarding the Kingma acceptance chain to Ilya.]\n\n> **From:** Diederik Kingma <kingma@…>\n> **To:** Greg Brockman <gdb@openai.com>\n> **Cc:** Sam Altman <sam@ycombinator.com>, John Schulman <joschu@openai.com>\n> **Subject:** Decision\n> **Date:** Sunday, December 6, 2015\n>\n> Hi, :-)\n>\n> I accept! :-)\n>\n> You might wonder what made me change my mind. First, perhaps most importantly, I realised that doing AI research the way OpenAI intends to, is almost certainly the more ethical choice in the long run. Second, fundamental AI research is simply much more interesting in the long run, and I really enjoy research work. Third, contributing to fundamental AI research will have the greatest expected long-term impact. Also, from my earlier e-mail: you have assembled an extraordinary initial team (and being part of the founding team has some advantages), I think the environment is going to be really special, and **Elon's involvement is awesome. It's an honour to be part of this team**. And, of course, the weather is a big step up ;-)\n>\n> Together, these things have made me very excited to join your lab. It's hard to describe how much I'm looking forward to it.\n>\n> D\n\n> **From:** Greg Brockman <gdb@openai.com>\n> **To:** Diederik Kingma <kingma@…>\n> **Subject:** Re: Decision\n> **Date:** Sat, Dec 5, 2015 at 7:22 PM\n>\n> Fantastic news!!!\n>\n> - I've attached a letter with an updated date of Dec 8.\n> - we're flexible on start date. Starting after graduating from your PhD in Summer 2016 works just fine. (We'll of course be happy for you to be as involved as you like prior to then, and could have you on the email lists, etc..)\n>\n> - gdb\n\n> **From:** Diederik Kingma <kingma@…>\n> **To:** Greg Brockman <gdb@openai.com>\n> **Subject:** Re: Decision\n> **Date:** Sat, Dec 5, 2015 at 6:17 AM\n>\n> Hi Greg,\n>\n> Makes sense.\n>\n> Some details before I sign:\n> - The offer letter has deadline Dec 1st. Not sure if it's legally valid if I accept that version.\n> - Regarding my start date: I can most probably finish my PhD in summer 2016. Please let me know what your constraints are regarding the start date.\n>\n> Regards, Durk\n\n> **From:** Greg Brockman <gdb@openai.com>\n> **To:** Diederik Kingma <kingma@…>\n> **Subject:** Re: Decision\n> **Date:** Thu, Dec 3, 2015 at 6:55 PM\n>\n> Hey Durk,\n>\n> Yep, makes sense.\n>\n> To add to what John said, I strongly agree with your statements. (B) is actually the core reason we're starting this lab. Even more strongly, on the world's current trajectory, AGI seems most likely to come out of a for-profit company, or maybe a government intelligence agency as part of an arms race. We have an opportunity to set the stage so that doesn't happen: I think it should happen as an international coalition, more like the ISS than the space race.\n>\n> The lab's mission is ensuring that AGI is beneficial, and that those benefits are distributed widely to the world rather than making anyone into a quadrillion-dollar company or omnipotent surveillance state. I don't care if we're the ones that ultimately make AGI, but I care that we shape the direction of progress so that AGI ultimately happens in the right way.\n>\n> Achieving this requires being a leading research institution. … - Open up the dialog about safety in the community: right now, I'm pretty sure the majority AI researchers believe safety is worth starting to think about, but it's pretty hard to talk about openly due to the vocal minority. - Set the standard for how AI research should happen (not just for AGI, but also for reproducibility, etc), and how to actionably incorporate safety into the work. - Figure out how to communicate to the mainstream: I think most people see the costs of AI (a la Terminator) but don't know what the benefits would be. Maybe this requires something crazy like getting more movies like Her made.\n>\n> I'll note it's quite possible that this is the only time one could start a group like this — for example, if Google/DeepMind were to cement their talent monopoly over the next 5 years, it'd just be impossible to get anything new started. (Of course, there are many more people entering the field now, so perhaps their monopoly would break naturally — it's hard to know right now.)\n>\n> So I think that now is the right time. I think you are exactly the right person to help make this effort succeed. And I think you'd learn a ton from people like Elon, Sam, and Alan Kay — some of the best people in the world at getting things done.\n>\n> As a parting thought, here is a quote from Elon (from a private email, please keep confidential):\n>\n> \"**On the time allocation front, I've decided to dedicate whatever amount of my time that is materially useful, even though it will come at some cost to SpaceX and Tesla. If I really believe that this is potentially the biggest near-term existential threat, then action should follow belief.**\"\n>\n> Anyway, I get to NIPS on Sunday, and would be happy to chat whenever; also happy to schedule an Elon call — let me know what would be most helpful!\n>\n> - gdb\n\n> **From:** John Schulman <joschu@openai.com>\n> **To:** Diederik Kingma <kingma@…>\n> **Subject:** Re: Decision\n> **Date:** Wed, Dec 2, 2015 at 10:32 PM\n>\n> Hey Durk, I'm really excited to hear that you're still considering joining us. I think you're the top person in the world in your field — certainly, you've written my favorite papers on generative models and semisupervised learning — so it would be sad if we weren't able to use your hard-earned mastery of these topics to make the next breakthroughs. And if you join us, you'd help us build our research program and culture so we can get the next generation of researchers up to speed, to build on your ideas in various directions.\n>\n> … Regarding human-level AI, one hard part of this discussion is that we don't really know what we want or what would be a good outcome (at least that's how I feel.) But you've hit the nail on the head that the best thing we can do right now is try to make it happen in a place where people have the right intentions and values that are consistent with ours. This lab will be unique as a top-notch research group where we mutually agree it is worth taking this issue seriously and talking about it candidly, so that as we understand it better, we can try to steer things in the right direction.\n>\n> John\n\n> **From:** Diederik Kingma <kingma@…>\n> **To:** Greg Brockman <gdb@openai.com>, John Schulman <joschu@openai.com>\n> **Subject:** Re: Decision\n> **Date:** Wed, Dec 2, 2015 at 4:29 PM\n>\n> P.S. Regarding the ethical aspect, I gave it a bit more thought this week, and my current take on it is pasted below. TL;DR: it's actually not unethical to contribute to AI research, as long as it's at the right place. … from an ethical viewpoint, AI researchers shouldn't be concerned by (A) [speeding up powerful AI]; rather, career decisions should be guided to a much greater degree by (B): whether it's developed in the right environment: a civilised western institution with some oversight is a much better environment than some close, obscure, foreign institution without oversight.\n\n> **From:** Greg Brockman <gdb@openai.com>\n> **To:** Diederik Kingma <kingma@…>\n> **Subject:** Re: Decision\n> **Date:** Sun, Nov 29, 2015 at 6:45 AM\n>\n> Hey Durk,\n>\n> I spent some time thinking about this, and was wondering: would you be up for chatting with Elon before you consider your decision final (and presumably after your voice returns :))?\n>\n> I think it'd be interesting for you to hear why he's decided to get more involved, and what he thinks is necessary for the world to have a good outcome from AI. I think he's probably the person with the best perspective in the world on how to actually achieve great outcomes on some of humanity's hardest problems …\n\n> **From:** Diederik Kingma <kingma@…>\n> **To:** Greg Brockman <gdb@openai.com>\n> **Subject:** Re: Decision\n> **Date:** Sat, Nov 28, 2015 at 2:02 PM\n>\n> Hi Greg,\n>\n> Thanks. I must say that part of me is sad as well.\n>\n> Regarding Skype: yesterday I completely lost my voice due to a cold, and Skype is infeasible. I just read that it usually takes at least a couple of days to heal. I'll let you know as soon as my voice recovers!\n>\n> Regards\n\n> **From:** Greg Brockman <gdb@openai.com>\n> **To:** Diederik Kingma <kingma@…>\n> **Subject:** Re: Decision\n> **Date:** Sat, Nov 28, 2015 at 7:37 PM\n>\n> Aww, definitely sad to hear it, though I'm excited for you. I'm certainly happy to be as useful as I can.\n>\n> And sure, I'd love to chat on Skype — I'm around the rest of today, on a plane tomorrow, and should be free Monday. When's best?\n>\n> - gdb\n\n> **From:** Diederik Kingma <kingma@…>\n> **To:** Greg Brockman <gdb@openai.com>, Sam Altman <sam@ycombinator.com>\n> **Subject:** Decision\n> **Date:** Sat, Nov 28, 2015 at 6:02 AM\n>\n> Hi Greg, Sam,\n>\n> This was not an easy decision, but after much deliberation, I decided to go with the automated radiology startup.\n>\n> You have assembled an extraordinary initial team, I think the environment is going to be really special. Of course, working with Elon would be awesome. And of course being part of the founding team comes with a big set of advantages. I really enjoy research work. I'm also honored to be asked for the position. All this together made it a tough decision. Sorry for it taking so long. In the end, the lure and potential of the startup persisted.\n>\n> As I told you, I do plan to keep publishing papers. Also I want to return to go back to do pure AI research eventually. In that case, I'll of course come back begging for a position in your lab :)\n>\n> If you want to do Skype call, please let me know.\n>\n> Regards, Durk\n>\n> P.S. We plan to incorporate in SF next year, and to do a financing round in Q1/Q2 of 2016. However our funding network isn't great in SF. You or Sam might be able to open some doors for us?\n>\n> Some info: one of the co-founders is another top ML researcher, Tim Salimans, who multiple prizes at Kaggle. He is also a well known figure in the Bayesian statistics community, and winner of the prestigious Lindley prize. The others are a biz dev and CS guy, both excellent. Tim just finished a prototype that outperforms radiologists on classifying arthritis in the knee. On the business side is going well too. We would gladly give a talk/presentation when we're in SF to anyone that is interested.\n>\n> --\n> Sent from mobile\n\n## Commentary\n\nPX 66 is a contemporaneous receipt of the founding-pitch language plaintiffs lean on. Three things about it travel well to a jury: (i) Diederik Kingma — a top deep-learning researcher who had already accepted a startup offer — flips back to OpenAI specifically because of \"Elon's involvement\"; (ii) Brockman, in his own recruiting prose, says OpenAI's mission is \"ensuring that AGI is beneficial, and that those benefits are distributed widely … rather than making anyone into a **quadrillion-dollar company**\"; and (iii) Brockman, with Musk's permission, quotes a private Musk email — the **\"biggest near-term existential threat … action should follow belief\"** line — that becomes, in plaintiffs' framing, a 2015 baseline against which the \"[[Brockman Journal|It would be nice to be making the billions]]\" 2017 journal entry ([[PX 151]] / [[Brockman Journal]]) is to be measured. Brockman's Day 6 cross by Mr. Kry put PX 66 at ~09:55 PT to set up the [[Key Themes|\"Greg Brockman's contributions ('zero')\"]] line of questioning that culminated in the $30B / paid-zero exchange.\n\n---\n*See also:* [[Day 6|Day 6 digest]] · [[Greg Brockman]] · [[Ilya Sutskever]] · [[Elon Musk]] · [[Brockman Journal]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-67", "exhibit": "PX 67", "party": "Plaintiffs", "type": "Email", "admitted_trial_day": "Day 6 (May 4, 2026)", "uploaded_box_pt": "2026-05-04T14:46:06", "uploader": "Someone", "pages": 1, "size_bytes": 365730, "source_pdf": "PX-67.pdf", "pdf_url": "https://media.mts-in.com/PX-67.pdf", "body_markdown": "# PX 67 — Dec 11, 2015 Musk \"Congratulations on a great beginning!\" launch email\n\n> Musk's morale-boosting note to the OpenAI founding technical team three days after incorporation — \"We are outmanned and outgunned by a ridiculous margin... but we have right on our side.\"\n\n## Document type\n**Email, single message, plain text.** From Elon Musk (erm@spacex.com) to the OpenAI founding research staff with Sam Altman cc'd. Bates 2024MUSK-0009787.\n\n## Logistics\n- **Trial admission:** Day 6 (May 4, 2026) — Russell direct + cross; Brockman cross by Mr. Kry.\n- **Box upload:** 2026-05-04 14:46:06 PT — Day 6 mid-afternoon batch (clustered with PX 154 and PX 151 from the same upload window).\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~357 KB, 1 page.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `067.pdf`.\n\n## Transcribed text\n\n> **From:** \"Elon Musk\" <erm@spacex.com>\n> **To:** Ilya Sutskever, Pamela Vagata, Vicki Cheung, Durk Kingma, Andrej Karpathy, John Schulman, tlb@ycombinator.com, Greg Brockman, Wojciech Zaremba\n> **Cc:** \"sam@ycombinator.com\" <sam@ycombinator.com>\n> **Subject:** The OpenAI Company\n> **Date:** Fri, 11 Dec 2015 16:41:18 -0800\n> **Importance:** Normal\n>\n> Congratulations on a great beginning!\n>\n> We are outmanned and outgunned by a ridiculous margin by organizations you know well, but we have right on our side and that counts for a lot. I like the odds.\n>\n> Our most important consideration is recruitment of the best people. The output of any company is the vector sum of the people within it. If we are able to attract the most talented people over time and our direction is correctly aligned, then OpenAI will prevail.\n>\n> To this end, please give a lot of thought to who should join. If I can be helpful with recruitment or anything else, I am at your disposal. I would recommend paying close attention to people who haven't completed their grad or even undergrad, but are obviously brilliant. Better to have them join before they achieve a breakthrough.\n>\n> Looking forward to working together,\n> Elon\n\n## Commentary\n\nPX 67 is plaintiffs' \"founding moment\" exhibit — Musk addressing the assembled founding researchers (Sutskever, Karpathy, Schulman, Zaremba, Brockman, Kingma, Cheung, Vagata) three days after the [[PX-16|Dec 8, 2015 incorporation]] and one day after the public launch. Plaintiffs use it for the **\"Without me, OpenAI wouldn't exist\"** point Musk made on direct (see [[Key Themes]] — the \"OPEN\" / open-source thread): the recruits were addressed by Musk, who paid them, and who positioned the venture against \"organizations you know well\" (Google/DeepMind). The recruiting line about \"people who haven't completed their grad or even undergrad\" anticipates the later cross-examination disputes about who built OpenAI's people pipeline. Read alongside [[PX-154|Brockman's contemporaneous journal]] from the same period — and Brockman's later \"**This is the only chance we have to get out from under Elon**\" passage in [[PX 151]] — this email frames the contrast plaintiffs press: the founders' public posture toward their patron versus what they were writing in private.\n\n---\n*See also:* [[Day 6|Day 6 digest]] · [[PX-16]] · [[PX 151]] · [[PX-154]] · [[Greg Brockman]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-7", "exhibit": "PX 7", "party": "Plaintiffs", "type": "Email thread", "admitted_trial_day": "Day 2 (April 28, 2026)", "uploaded_box_pt": "2026-04-28T14:51:57", "uploader": "Someone", "pages": 2, "size_bytes": 242367, "source_pdf": "PX-7.pdf", "pdf_url": "https://media.mts-in.com/PX-7.pdf", "body_markdown": "# PX 7 — Oct 18–19, 2015 Altman → Musk \"followup\" email (the $100MM ask)\n\n> The pre-incorporation Altman → Musk email setting out the original three \"specific asks\" — a $30MM five-year donation, an \"advisor\" title, and a Safety Board seat — atop Musk's brief reply about governance.\n\n## Document type\n**Email thread, plain text, two messages.** Sam Altman's October 19, 2015 8:47 AM PDT reply on top of his October 18, 2015 11:36 AM original to Musk, with Musk's intervening October 19, 2015 3:27 AM \"Let's discuss governance\" reply quoted in the middle. Subject \"Re: followup.\" Bates 2024MUSK-0005439–5440.\n\n## Logistics\n- **Trial admission:** Day 2 (April 28, 2026) — used during plaintiffs' opening / Musk direct as the \"founding ask\" exhibit. No prior wiki reference; admission day inferred from Box upload date and Day 2 cluster.\n- **Box upload:** 2026-04-28 14:51:57 PT — Day 2 mid-afternoon batch.\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~237 KB, 2 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `007.pdf`.\n\n## Transcribed text\n\n> **From:** Sam Altman <sam@ycombinator.com>\n> **To:** Elon Musk <erm@spacex.com>\n> **Subject:** Re: followup\n> **Date:** Mon, 19 Oct 2015 08:47:19 -0700\n> **Importance:** Normal\n>\n> Happy to talk about it any time--very focused on getting this right!\n>\n> Thanks--you were actually the main inspiration for that. We've had such a struggle getting our capital-intensive businesses funded (unless the founders were already rich from a previous startup) that we figured we should just do it ourselves.\n\n> **From:** Elon Musk <erm@spacex.com>\n> **To:** Sam Altman <sam@ycombinator.com>\n> **Subject:** Re: followup\n> **Date:** Mon, Oct 19, 2015 at 3:27 AM\n>\n> Let's discuss governance. This is critical. I don't want to fund something that goes in what turns out to be the wrong direction.\n>\n> Btw, great move on the continuity fund!\n\n> **From:** Sam Altman <sam@ycombinator.com>\n> **To:** Elon Musk <erm@spacex.com>\n> **Subject:** followup\n> **Date:** Sun, Oct 18, 2015 at 11:36 AM\n>\n> Good seeing you a couple of weeks ago.\n>\n> As discussed I think starting with a $100MM commitment (and leaving the time unspecified) is the way to go.\n>\n> Everyone we discussed but Gates is committed to donating. Hope to have him locked down next week. I am going to sit down with Zuck, but I still think it's probably too problematic given Facebook AI Research.\n>\n> Two big new recruits that we expect to sign offer letters this week: John Schulman and Wojciech Zaremba.\n>\n> Specific asks for you:\n>\n> 1) Can you donate $30MM over the next 5 years?\n>\n> 2) Can we call you an \"advisor\" but leave unspecified what you'll do in detail and figure it out as we go?\n>\n> 3) Will you be on the Safety Board with me? I'd like it to eventually be 5 of us, adding the next 3 over the year or so. This will be the \"second key\" for releasing anything that could be dangerous.\n>\n> Sam\n\n## Commentary\n\nPX 7 is a foundational plaintiffs' exhibit. It establishes: (i) that Musk's role at OpenAI was conditioned from day one on **governance** (\"This is critical. I don't want to fund something that goes in what turns out to be the wrong direction\"); (ii) the original $100MM target and Musk's $30MM five-year ask — relevant to the [[$1B / $38M gap]] dispute the defense pressed; and (iii) the \"Safety Board\" / \"second key\" concept Altman described as \"for releasing anything that could be dangerous,\" which plaintiffs use to anchor their open-source / safety-mission framing. Musk's \"great move on the continuity fund!\" line is the defense's counterweight: praise for YC's *for-profit* continuity fund, weeks before OpenAI's nonprofit launch. Cross-reference: this email pre-dates the December 11, 2015 [[PX 5|OpenAI launch blog post]] and the [[PX 24]] Charter language by ~8 weeks.\n\n---\n*See also:* [[Day 2|Day 2 digest]] · [[Sam Altman]] · [[PX 5]] · [[PX 24]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-70", "exhibit": "PX 70", "party": "Plaintiffs", "type": "Email thread", "admitted_trial_day": "Day 2 (April 28, 2026)", "uploaded_box_pt": "2026-04-28T14:51:58", "uploader": "Someone", "pages": 1, "size_bytes": 197889, "source_pdf": "PX-70.pdf", "bates": "2024MUSK-0004103", "pdf_url": "https://media.mts-in.com/PX-70.pdf", "body_markdown": "# PX 70 — Musk–Huang \"DGX-1\" / \"good intentions\" email (April 13–20, 2016)\n\n> Musk's April 13, 2016 email to Nvidia CEO Jensen Huang asking whether OpenAI can buy one of the early DGX-1 supercomputers, describing OpenAI as \"a non-profit funded by me and a few others with the goal of developing safe AGI (and hopefully not paving the road to hell with good intentions).\"\n\n## Document type\n**Email thread, plain text, three messages on one page.** Bottom: Musk to Huang, April 13, 2016 12:01 AM. Middle: Huang's April 19, 2016 10:11 AM reply. Top: Musk's April 20, 2016 8:20:42 AM \"Much appreciated.\" Marked Highly Confidential-AEO. Bates 2024MUSK-0004103.\n\n## Logistics\n- **Trial admission:** Day 2 (April 28, 2026) — used during Musk direct examination as part of plaintiffs' opening narrative on Musk's 2016 commitment to OpenAI as a nonprofit. Per [[Key Themes]] this is referred to as \"PX 70 — Musk-Huang email April 2016 (good intentions).\"\n- **Box upload:** 2026-04-28 14:51:58 PT — Day 2 mid-afternoon batch (the openings + Musk-direct day's first wave of exhibits).\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~198 KB, 1 page.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `070.pdf`.\n\n## Transcribed text\n\n> **From:** Elon Musk <erm@tesla.com>\n> **To:** Jensen H Huang <[redacted]@nvidia.com>\n> **Subject:** Re: DGX-1\n> **Date:** 4/20/2016 8:20:42 AM\n>\n> Much appreciated\n\n> **From:** Jensen H Huang <[redacted]@nvidia.com>\n> **To:** Elon Musk <erm@tesla.com>\n> **Subject:** Re: DGX-1\n> **Date:** Tue, Apr 19, 2016 at 10:11 AM\n>\n> I've not forgotten.\n>\n> This is the first supercomputer that's selling itself off the web! Demand coming from all over.\n>\n> First shipment starts end next month. And already taking orders for delivery in September.\n>\n> I will make sure OpenAI gets one of the first ones.\n\n> **From:** Elon Musk <erm@tesla.com>\n> **To:** Jensen H Huang <[redacted]@nvidia.com>\n> **Subject:** DGX-1\n> **Date:** Wednesday, April 13, 2016 12:01 AM\n>\n> Can OpenAI buy one of the early units? The team was asking me about this earlier today.\n>\n> OpenAI is unaffiliated with Tesla. It is a non-profit funded by me and a few others with the goal of developing safe AGI (and hopefully not paving the road to hell with good intentions).\n\n## Commentary\n\nPX 70 is one of plaintiffs' two cleanest contemporaneous statements by Musk himself describing OpenAI to a third party as \"a non-profit … developing safe AGI\" — and explicitly distinguishing it from Tesla (\"OpenAI is unaffiliated with Tesla\"). Plaintiffs use it to undercut the defense's \"tale of two Elons\" framing: in April 2016, before any of the 2017 control fights or 2018 board exit, Musk was telling Nvidia's CEO in plain text that OpenAI was a non-profit and that he was funding it. The \"paving the road to hell with good intentions\" line — Musk's own gallows-humor phrasing — is also the line that gives the exhibit its informal name in [[Key Themes]]. The DGX-1 itself is a recurring trial motif: Nvidia's first deep-learning supercomputer, hand-delivered to OpenAI later in 2016 (the famous \"first DGX-1\" photograph), and the seed of the compute-asymmetry story that plaintiffs build into their breach-of-trust theory by 2019. Compare with [[PX 17]] (the December 2015 launch blog post) for the public-facing version of the same nonprofit framing.\n\n---\n*See also:* [[Day 2|Day 2 digest]] · [[PX 17]] · [[PX 156]] · [[PX 157]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-71", "exhibit": "PX 71", "party": "Plaintiffs", "type": "Email thread with attachments (fiscal sponsorship)", "admitted_trial_day": "Day 4 (April 30, 2026)", "uploaded_box_pt": "2026-04-30T15:27:07", "uploader": "Someone", "pages": 12, "size_bytes": 3785778, "source_pdf": "PX-71.pdf", "pdf_url": "https://media.mts-in.com/PX-71.pdf", "body_markdown": "# PX 71 — Feb–May 2016 fiscal-sponsorship email thread (YC.org → OpenAI; Articles, IRS Letter, Sponsor Letter)\n\n> The mechanical paperwork showing how Musk's quarterly $5M wires actually reached OpenAI in 2016 — through YC.org as fiscal sponsor — including YC.org's articles of incorporation, its IRS 501(c)(3) determination letter, and the May 2016 sponsor letter asserting \"variance power\" over the funds.\n\n## Document type\n**Email thread plus attached corporate/tax records.** Twelve-page compilation: a long Chris Clark (then COO of OpenAI; later YC.org Treasurer) ↔ Ron Gong (CTC|myCFO/BMO) email thread Feb–May 2016 setting up the fiscal-sponsorship plumbing for Musk Foundation contributions; followed by the YC.org Articles of Incorporation (filed Apr 3, 2015), the YC Inc. naming-consent letter (3/27/2015), the IRS Section 501(c)(3) determination letter for YC.org (effective Apr 3, 2015, EIN 47-3772053), and a May 20, 2016 letter from Chris Clark to Ron Gong confirming YC.org as fiscal sponsor of \"The OpenAI Artificial Intelligence Research Program\" with \"variance power\" over donated funds. Bates BMO_00000120–131.\n\n## Logistics\n- **Trial admission:** Day 4 (April 30, 2026) — used during Birchall direct on the mechanics of Musk's contributions and (with [[PX 87]]) to support the plaintiffs' fiscal-sponsorship/donor-advised-fund theory of the donation chain. No prior wiki reference; admission day inferred from Box upload date.\n- **Box upload:** 2026-04-30 15:27:07 PT — late-afternoon Day 4 batch.\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~3.6 MB, 12 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `071.pdf`.\n\n## Transcribed text\n\n> **From:** Chris Clark <chris@ycr.org>\n> **To:** Ronald Gong <[redacted]@ctcmycfo.com>\n> **Cc:** Harris myCFO MUS001FOS, Jared Birchall <jbirchall@muskfoundation.org>, Jonathan Levy <jlevy@ycombinator.com>\n> **Subject:** Re: cash\n> **Date:** 5/21/2016 12:08:56 AM\n> **Attachments:** YC.org Incorporation Docs.pdf; YC.org 501(c)(3) IRS Determination Letter.pdf; YC.org (Sponsor) Letter OpenAI Donors May 2016.pdf; yc.org svb wire instructions.pdf\n>\n> Ron,\n>\n> I have some good news to share on our end. The fiscal sponsorship arrangement I'd previously described is being executed today, and YC.org is ready to accept donations. I have attached the following:\n>\n> 1) YC.org articles of incorporation\n> 2) YC.org IRS determination letter\n> 3) Letter describing the sponsorship arrangement\n> 4) Wire transfer instructions\n>\n> YC.org will be able to grant funds to OpenAI via the sponsorship arrangement described in the letter.\n>\n> I know we're a few months behind, but we're hoping you'll be able to make the Q2 donation now for April-June, and then we can start a regular quarterly donation schedule starting in early July for Q3.\n>\n> Please let me know if I can answer any additional questions, and thanks again for your patience over the last several weeks.\n>\n> Thanks,\n> Chris Clark\n\n> **From:** Chris Clark <chris@ycr.org>\n> **To:** Ronald Gong <[redacted]@ctcmycfo.com>\n> **Subject:** Re: cash\n> **Date:** Thu, Apr 28, 2016 at 12:46 AM\n>\n> Hi Ron,\n>\n> Yes, we are in the process of taking the necessary board actions to put the fiscal sponsorship in place. YC.org, which has received a determination letter, will be the entity to which the donations are made, and YC.org will then be able to issue project grants to OpenAI. We will provide a letter explaining everything for your records.\n>\n> This should all be setup early next week. I'm just waiting for the YC.org board to take the necessary actions.\n>\n> Chris\n\n> **From:** Ronald Gong <[redacted]@ctcmycfo.com>\n> **To:** Chris Clark <chris@ycr.org>\n> **Subject:** Re: cash\n> **Date:** Wed, Apr 27, 2016 at 3:43 PM\n>\n> HI Chris – just checking back here. Please let us know if there are updates.\n>\n> thx\n>\n> Ron Gong\n\n> **From:** Chris Clark <chris@ycr.org>\n> **To:** Ronald Gong <[redacted]@ctcmycfo.com>\n> **Cc:** Harris myCFO MUS001FOS, Jared Birchall\n> **Subject:** Re: cash\n> **Date:** Saturday, April 02, 2016 7:58 PM\n>\n> Hi Ron,\n>\n> Quick Update: We received good news about using YC.org as a fiscal sponsor for OpenAI, so we are putting that in motion next week and hope to be ready to accept funds the following week.\n>\n> I will send you all the info for YC.org next week. We already have a determination letter for that organization, which should hopefully simplify everything.\n>\n> Thanks,\n> Chris\n\n> **From:** Ronald Gong <[redacted]@ctcmycfo.com>\n> **To:** Chris Clark <chris@ycr.org>\n> **Subject:** Re: cash\n> **Date:** Mon, Mar 21, 2016 at 9:37 PM\n>\n> Thanks for the update, Chris. We're fine with any delay.\n>\n> Sent from my BlackBerry 10 smartphone.\n\n> **From:** Chris Clark <chris@ycr.org>\n> **To:** Ronald Gong <[redacted]@ctcmycfo.com>\n> **Cc:** Harris myCFO MUS001FOS, Jared Birchall\n> **Subject:** Re: cash\n> **Date:** Monday, March 21, 2016 9:21 PM\n>\n> Ron,\n>\n> I just want to update you on this since it's been a few days. We are trying to determine whether we can use one of our existing 501(c) orgs as a fiscal sponsor instead of going through a 3rd party since the 3rd party sponsor would take a sizable sum off the top. I'm chatting again with the legal team on Weds.\n>\n> We will sort everything out as soon as possible, but worst case scenario is that we request a delay for the April 1st donation while we get this settled. Hopefully it won't be too much trouble to send this quarter's funds a little later in April and then resume a regular quarterly schedule after that, but I will do my best to still meet the April 1st date.\n>\n> Thanks,\n> Chris Clark\n> Chief Operating Officer\n> OpenAI\n\n> **From:** Ronald Gong <[redacted]@ctcmycfo.com>\n> **To:** Chris Clark <chris@ycr.org>\n> **Subject:** Re: cash\n> **Date:** Fri, Mar 11, 2016 at 3:12 PM\n>\n> Thanks for the update Chris.\n>\n> Sent from my BlackBerry 10 smartphone.\n\n> **From:** Chris Clark <chris@ycr.org>\n> **To:** Ronald Gong <[redacted]@ctcmycfo.com>\n> **Cc:** Harris myCFO MUS001FOS, Jared Birchall, chris@ycr.org\n> **Subject:** Re: cash\n> **Date:** Friday, March 11, 2016 1:39 PM\n>\n> Hi Ron,\n>\n> Thanks for checking in. I'm waiting for the legal team to get me some additional info. I have everything except the 1023.\n>\n> OpenAI hasn't received a determination letter yet, so I'm also getting the status of fiscal sponsorship from them.\n>\n> I'll get back to you asap.\n>\n> Chris\n\n> **From:** Ronald Gong <[redacted]@ctcmycfo.com>\n> **To:** Chris Clark <chris@ycr.org>\n> **Subject:** Re: cash\n> **Date:** Fri, Mar 11, 2016 at 9:47 AM\n>\n> Hi Chris -- checking back. How is this coming along?\n\n> **From:** Chris Clark <chris@ycr.org>\n> **To:** Ronald Gong <[redacted]@ctcmycfo.com>\n> **Subject:** Re: cash\n> **Date:** Wednesday, March 02, 2016 1:27 AM\n>\n> Hi Ron,\n>\n> I'll get this info to you as soon as I have it.\n>\n> Thanks,\n> Chris\n\n> **From:** Sam Altman <sam@ycombinator.com>\n> **To:** Ronald Gong <[redacted]@ctcmycfo.com>\n> **Subject:** Re: cash\n> **Date:** Tue, Mar 1, 2016 at 10:51 AM\n>\n> Chris Clark starts next week and can provide all this info then -- thanks!\n>\n> Sam\n\n> **From:** Ronald Gong <[redacted]@ctcmycfo.com>\n> **To:** Sam Altman <sam@ycombinator.com>\n> **Subject:** Re: cash\n> **Date:** Mon, Feb 29, 2016 at 3:50 PM\n>\n> Hi Sam – great to speak to you last week.\n>\n> Regarding the entity's pending tax exempt status, can you provide the following:\n>\n> - A copy of the Form 1023 filed with the IRS while application for tax exemption status is pending;\n> - Entity's name, address and federal tax ID# (most likely already on the Form 1023)\n> - Once receives the tax exempt status, please forward the IRS Exemption Determination Letter\n>\n> Best,\n> Ron\n\n> **From:** Elon Musk <erm@spacex.com>\n> **To:** Sam Altman <sam@ycombinator.com>, Ronald Gong <[redacted]@ctcmycfo.com>\n> **Subject:** Re: cash\n> **Date:** Thursday, February 25, 2016 6:36 PM\n>\n> Ron, we need to wire $5M per quarter to OpenAI, starting April 1. This is a non-profit.\n\n> **From:** Sam Altman <sam@ycombinator.com>\n> **To:** Elon Musk <erm@spacex.com>\n> **Subject:** Re: cash\n> **Date:** Thu, Feb 25, 2016 at 6:32 PM\n>\n> ok thanks, do you have an assistant or finance person or something i can coordinate with?\n\n> **From:** Elon Musk <erm@spacex.com>\n> **To:** Sam Altman <sam@ycombinator.com>\n> **Subject:** Re: cash\n> **Date:** Thu, Feb 25, 2016 at 6:30 PM\n>\n> Agreed\n>\n> Yes\n\n> **From:** Sam Altman <sam@ycombinator.com>\n> **To:** Elon Musk <erm@spacex.com>\n> **Subject:** cash\n> **Date:** Thu, Feb 25, 2016 at 6:13 PM\n>\n> I think we're going to need more than I was originally budgeting given a) the salaries in the field and b) the speed at which you want to grow.\n>\n> Can you do $20MM a year for the each of the next 3 years? I can do $10MM/year, and we'll have brought $5MM a year from other donors.\n>\n> Sam\n\n> **From the May 20, 2016 fiscal-sponsor letter (Chris Clark → Ron Gong):**\n>\n> \"This letter is to confirm to you that YC.org is the fiscal sponsor of a charitable program known as 'The OpenAI Artificial Intelligence Research Program' that is currently being conducted by OpenAI, Inc. YC.org is able to receive and hold donations received in charitable trust, restricted to the purposes of the Programs.\n>\n> YC.org is a California nonprofit public benefit corporation qualified as exempt from federal income tax under Section 501(c)(3) of the Internal Revenue Code ('IRC') and classified as a public charity under IRC Sections 509(a)(1) and 170(b)(1)(A)(vi)…\n>\n> For legal and accounting purposes, we need to notify you of YC.org's **'variance power'** over funds donated in support of the Program. Under the fiscal sponsorship agreement with OpenAI, Inc., YC.org retains full discretion and control over the use of funds to accomplish the charitable purposes of the approved program. **This power includes the unilateral right to redirect funds to a different beneficiary** who can accomplish the purposes of this program if for some reason OpenAI, Inc. cannot. (See Accounting Standards Codification paragraphs ASC 958-605-25-25 and -26.)\"\n\n> **From the YC.org Articles of Incorporation (filed April 3, 2015, signed by Leila M. Stevens, Incorporator, dated March 30, 2015):**\n>\n> \"II.A. This corporation is a nonprofit Public Benefit Corporation and is not organized for the private gain of any person. It is organized under the Nonprofit Public Benefit Corporation Law for public and charitable purposes.\n>\n> II.B. The specific purposes of this corporation are to support other charitable organizations conducting charitable activities and advance science and education through the distribution of research grants…\n>\n> VI. The property of this corporation is irrevocably dedicated to the purposes in Article II(a) hereof and no part of the net income or assets of this corporation shall ever inure to the benefit of any director, officer or member thereof or to the benefit of any private person…\"\n\n[Remainder: pages 10–11 are the IRS Letter 947 (Section 501(c)(3) determination, EIN 47-3772053, effective April 3, 2015, public charity under §170(b)(1)(A)(vi)) — boilerplate IRS exemption letter signed Jeffrey I. Cooper, Director, Exempt Organizations Rulings and Agreements; full doc available in source PDF.]\n\n## Commentary\n\nPX 71 is plaintiffs' \"show your work\" exhibit on the donation pipeline. It establishes three things their charitable-trust theory needs: (1) the **$5M-per-quarter cadence** Musk authorized in February 2016 for \"the next 3 years\" — so the contributions had a defined purpose and time horizon; (2) the **fiscal-sponsor structure** running through YC.org, a separately incorporated 501(c)(3) public charity whose articles dedicate property \"irrevocably\" to its charitable purposes; and (3) the **variance-power letter**, which legally subordinates the funds to charitable use and gives YC.org the unilateral right to redirect them away from OpenAI if OpenAI fails to carry out the program. Together these documents anchor plaintiffs' argument that the donations were not unrestricted gifts to a private corporation but charitable dollars held in trust for a defined research mission. (See also [[PX 87]], the YC.org $5M tax-acknowledgment letter dated May 26, 2017.) The defense reading is that Musk was on notice from day one that YC.org — not he, and not OpenAI Inc. directly — held discretion over the funds; that the variance-power language is standard ASC 958 fiscal-sponsorship boilerplate; and that a tax deduction was taken every step of the way (see [[Key Themes]] § \"donor-advised funds / tax-deduction wrinkle\"). Birchall confirmed on Day 4 that Musk took a tax deduction for every contribution.\n\n---\n*See also:* [[Day 4|Day 4 digest]] · [[PX 87]] · [[Jared Birchall]] · [[Sam Altman]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-73", "exhibit": "PX 73", "party": "Plaintiffs", "type": "Letter", "admitted_trial_day": "Day 4 (April 30, 2026)", "uploaded_box_pt": "2026-04-30T15:26:48", "uploader": "Someone", "pages": 1, "size_bytes": 192727, "source_pdf": "PX-73.pdf", "pdf_url": "https://media.mts-in.com/PX-73.pdf", "body_markdown": "# PX 73 — May 27, 2016 YC.org $500,000 gift acknowledgment letter to Musk\n\n> A one-page tax-acknowledgment letter from YC.org Treasurer Chris Clark to Elon Musk recording a $500,000 gift on May 27, 2016 — the kind of \"no goods or services\" 501(c)(3) receipt that supported a federal charitable deduction.\n\n## Document type\n**Letter (charitable contribution acknowledgment).** One page on YC.org letterhead, signed by Chris Clark, Treasurer, citing YC.org's EIN 47-3772053. Bates BMO_00000172 — a Birchall/Musk-foundation document set.\n\n## Logistics\n- **Trial admission:** Day 4 (April 30, 2026) — used during the Birchall direct/cross sequence, where plaintiffs traced the precise $38,191,066 in ~60 contributions and the defense pressed the [[donor-advised funds]] / tax-deduction wrinkle. No prior wiki reference; admission day inferred from Box upload date.\n- **Box upload:** 2026-04-30 15:26:48 PT — late-afternoon Day 4 batch.\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~188 KB, 1 page.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `073.pdf`.\n\n## Transcribed text\n\n> YC.org\n> 335 Pioneer Way\n> Mountain View, CA 94041\n>\n> May 27, 2016\n>\n> Dear Mr. Musk:\n>\n> This letter is to gratefully acknowledge, for your tax records, our receipt of your generous gift of $500,000 on May 27, 2016.\n>\n> No goods or services were provided to you in exchange for this gift. Therefore, the full amount of your gift qualifies as a charitable contribution for federal tax purposes. For reference, YC.org's EIN is: 47-3772053.\n>\n> Thank you again for your generous gift!\n>\n> Sincerely,\n>\n> /s/ Chris Clark\n> Chris Clark\n> Treasurer\n> YC.org\n\n## Commentary\n\nPX 73 is one of plaintiffs' building-block exhibits for the [[$1B / $38M gap]] dispute and the [[donor-advised funds]] tax-deduction wrinkle. The May 27, 2016 $500,000 contribution — routed through YC.org rather than directly to OpenAI Inc. — is part of the ~60-payment trail Birchall confirmed on direct (PX 112A: total $38,191,066). Wilson's cross use of letters like this one anchored the question, \"for every single one of those donations, it's somewhere in the chain, Mr. Musk was entitled to take a tax deduction?\" — Birchall: \"Yes, that's the nature of philanthropic giving.\" (See [[Key Themes]].) The \"no goods or services\" boilerplate is the legal tag that triggered the deduction; plaintiffs frame the deduction as confirmation Musk treated the money as charitable, while defense uses the same fact to argue Musk lost any claim of personal restitution once the funds passed through a 501(c)(3) channel.\n\n---\n*See also:* [[Day 4|Day 4 digest]] · [[Jared Birchall]] · [[PX 112A]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-74", "exhibit": "PX 74", "party": "Plaintiffs", "type": "Email thread", "admitted_trial_day": "Day 6 (May 4, 2026)", "uploaded_box_pt": "2026-05-04T14:46:06", "uploader": "Someone", "pages": 1, "size_bytes": 221724, "source_pdf": "PX-74.pdf", "pdf_url": "https://media.mts-in.com/PX-74.pdf", "body_markdown": "# PX 74 — May 30–31, 2016 Musk ↔ Brockman, \"uneasy feelings about Ilya\"\n\n> A short three-message thread that began as a routine O-1 visa-letter ask for OpenAI's first designer Ludwig Pettersson and pivots, in Musk's reply, to \"I'm having some uneasy feelings about Ilya btw. When I send you an email that doesn't copy him, that's intentional. We should talk about this privately.\"\n\n## Document type\n**Email thread, plain text, three messages** between Elon Musk (elon@openai.com) and Greg Brockman (gdb@openai.com), May 30–31, 2016. Bates OPENAI_MUSK00000114.\n\n## Logistics\n- **Trial admission:** Day 6 (May 4, 2026) — used during Mr. Kry's cross of Brockman to establish Musk's early-1H-2016 distrust of Ilya Sutskever and the Musk–Brockman side-channel that excluded the lead scientist Musk had recruited five months earlier.\n- **Box upload:** 2026-05-04 14:46:06 PT — Day-6 morning batch with the rest of the Brockman-cross set.\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~217 KB, 1 page.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `074.pdf`.\n\n## Transcribed text\n\n> **From:** Elon Musk <elon@openai.com>\n> **To:** Greg Brockman <gdb@openai.com>\n> **Subject:** Re: visa letter\n> **Date:** 5/31/2016 4:43:51 AM\n>\n> Great, if you know him that well, then that's good enough for me.\n\n> **From:** Greg Brockman <gdb@openai.com>\n> **To:** Elon Musk <elon@openai.com>\n> **Subject:** Re: visa letter\n> **Date:** Mon, May 30, 2016 at 9:42 PM\n>\n> Awesome, will send you the letter soon. He was the first designer at Stripe — single handedly took us from ugly to super well-designed — and was there for about 5 years. He does visual and UX design, and is definitely the best person I know at both.\n>\n> Sorry to hear that re: Ilya. Happy to chat at any time.\n>\n> - gdb\n\n> **From:** Elon Musk <elon@openai.com>\n> **To:** Greg Brockman <gdb@openai.com>\n> **Subject:** Re: visa letter\n> **Date:** Mon, May 30, 2016 at 9:25 PM\n>\n> I'm happy to sign any O-1 letters.\n>\n> What's his background?\n>\n> **I'm having some uneasy feelings about Ilya btw. When I send you an email that doesn't copy him, that's intentional.**\n>\n> **We should talk about this privately.**\n\n> **From:** Greg Brockman <gdb@openai.com>\n> **To:** Elon Musk <elon@openai.com>\n> **Subject:** visa letter\n> **Date:** Mon, May 30, 2016 at 4:49 PM\n>\n> Hi Elon,\n>\n> Ludwig Pettersson (our first designer — did gym.openai.com and has a lot more coming soon) is applying for an O-1 visa and was wondering if you'd be up for signing a reference letter for him.\n>\n> I'm sure you're familiar with how this works, but we'd provide the draft and there's no work beyond the signature. No worries if not!\n>\n> - gdb\n\n## Commentary\n\nPX 74 is a small but useful evidentiary brick for plaintiffs' \"control\" theme. It documents Musk in 2016 — well before the September 2017 final-straw fight — instructing Brockman to keep correspondence away from Sutskever and to \"talk about this privately.\" Plaintiffs use the exhibit to undermine the defense narrative that Musk's 2017 \"unequivocal control\" demand came out of nowhere; defense reads the same email as evidence of legitimate Musk concerns about a co-founder he had recruited from Google. The thread also sits chronologically in the same months as DX 539 (\"I don't want Sam on the lease\") and DX 600 (\"Elon holds 100 percent of the power and authority\") — see [[Key Themes]] §\"The Pioneer Building.\" It closes with Brockman's casual \"Sorry to hear that re: Ilya. Happy to chat at any time,\" confirming receipt of and willingness to accommodate the side-channel.\n\n---\n*See also:* [[Day 6|Day 6 digest]] · [[Greg Brockman]] · [[Ilya Sutskever]] · [[Elon Musk]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-75", "exhibit": "PX 75", "party": "Plaintiffs", "type": "Email thread", "admitted_trial_day": "Day 4 (April 30, 2026)", "uploaded_box_pt": "2026-04-30T15:27:16", "uploader": "Someone", "pages": 12, "size_bytes": 4475996, "source_pdf": "PX-75.pdf", "pdf_url": "https://media.mts-in.com/PX-75.pdf", "body_markdown": "# PX 75 — June 3–13, 2016 Birchall ↔ Jariwala Pioneer Building lease email thread\n\n> Twelve-page email thread between Jared Birchall (Musk family office) and Atit Jariwala (Bridgeton Holdings, the landlord) negotiating the Pioneer Building lease for OpenAI — including Musk's directive to remove Sam Altman's name from the lease and the wiring instructions for the first month's $141,666.67 rent payment to Bridgeton Pioneer Property, LLC.\n\n## Document type\n**Email thread, 12 pages.** Forwarded chain originating June 3, 2016 (\"Please review\" — Musk to Birchall, attaching Jariwala's lease draft) through final wiring instructions June 13, 2016. Bates BMO_00000192–0000203. Stamped \"Confidential.\" Phone numbers throughout redacted in black overlays. Signed \"Jared J. Birchall, Managing Director, Musk Family Office.\"\n\n## Logistics\n- **Trial admission:** Day 4 (April 30, 2026). Used during Birchall direct/cross to develop the [[Key Themes]] § \"Pioneer Building\" chain of custody: that the building was negotiated, signed, and paid for from the Musk family office, with Sam Altman explicitly removed from the lease at Musk's instruction.\n- **Box upload:** 2026-04-30 15:27:16 PT — late-afternoon Day 4 batch (clustered with PX 52, PX 87, PX 92).\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~4.4 MB, 12 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `075.pdf`.\n\n## Transcribed text\n\n> **From:** Jared Birchall <jbirchall@muskfoundation.org>\n> **To:** Harris myCFO MUS001FOS <MUS001FOS@ctcmycfo.com>\n> **Subject:** Fwd: Pioneer Building Open AI lease comments\n> **Date:** 6/13/2016 6:53:33 PM\n>\n> Below are the instructions for the aforementioned 1st months's lease payment. The amount is $141,666.67.\n>\n> *Jared J. Birchall, Managing Director, Musk Family Office*\n\n> **From:** Bourke Lee <blee@bridgetonholdings.com>\n> **To:** Jared Birchall <jbirchall@muskfoundation.org>\n> **Cc:** Atit Jariwala <ajariwala@bridgetonholdings.com>\n> **Subject:** RE: Pioneer Building Open AI lease comments\n> **Date:** Mon, Jun 13, 2016 at 12:25 PM\n>\n> Hi Jared,\n>\n> Nice to meet you over email. Please see the instructions below:\n>\n> Account Name: **Bridgeton Pioneer Property, LLC**\n> Account #: [redacted]\n> ABA #: 021001033\n> Deutsche Bank Trust Company Americas\n> New York, NY\n>\n> Thanks,\n> Bourke\n\n> **From:** Jared Birchall <jbirchall@muskfoundation.org>\n> **To:** Atit Jariwala <ajariwala@bridgetonholdings.com>\n> **Cc:** Bourke Lee <blee@bridgetonholdings.com>\n> **Subject:** Re: Pioneer Building Open AI lease comments\n> **Date:** Monday, June 13, 2016 2:23 PM\n>\n> Got it, thanks! And I'm pending the wiring instruction from lee to send the first month's rent. We'll get it sent asap.\n\n> **From:** Atit Jariwala <ajariwala@bridgetonholdings.com>\n> **To:** Jared Birchall <jbirchall@muskfoundation.org>\n> **Subject:** Re: Pioneer Building Open AI lease comments\n> **Date:** Mon, Jun 13, 2016 at 12:19 PM\n>\n> Yes, to Bridgeton Pioneer Fee, LLC. (the fee stands for \"fee interest\")\n\n> **From:** Jared Birchall <jbirchall@muskfoundation.org>\n> **To:** Atit Jariwala <ajariwala@bridgetonholdings.com>\n> **Cc:** Bourke Lee <blee@bridgetonholdings.com>\n> **Subject:** Re: Pioneer Building Open AI lease comments\n> **Date:** Monday, June 13, 2016 2:17 PM\n>\n> Got it, thanks. To confirm, Bridgeton Pioneer Fee, LLC? Not fund?\n\n> **From:** Atit Jariwala <ajariwala@bridgetonholdings.com>\n> **To:** Jared Birchall <jbirchall@muskfoundation.org>\n> **Subject:** Re: Pioneer Building Open AI lease comments\n> **Date:** Mon, Jun 13, 2016 at 12:15 PM\n>\n> Hi Jared,\n>\n> The beneficiary for the letter of credit should actually be: Bridgeton Pioneer Fee, LLC (this is different than from the monthly rent payments, which are to Bridgeton Pioneer Property LLC).\n>\n> Thanks,\n> Atit\n\n> **From:** Jared Birchall <jbirchall@muskfoundation.org>\n> **To:** Atit Jariwala <ajariwala@bridgetonholdings.com>\n> **Cc:** Bourke Lee <blee@bridgetonholdings.com>\n> **Subject:** Re: Pioneer Building Open AI lease comments\n> **Date:** Monday, June 13, 2016 12:47 PM\n>\n> Thanks! And who is the beneficiary of the letter of credit? Bridgeton Holdings? Or your lender?\n\n> **From:** Atit Jariwala <ajariwala@bridgetonholdings.com>\n> **To:** Jared Birchall <jbirchall@muskfoundation.org>\n> **Subject:** Re: Pioneer Building Open AI lease comments\n> **Date:** Monday, June 13, 2016\n>\n> That sounds good. Copying Bourke in my office who will provide the wiring instructions for the Pioneer account.\n>\n> Thanks, Jared.\n\n> **From:** Jared Birchall <jbirchall@muskfoundation.org>\n> **To:** Atit Jariwala <ajariwala@bridgetonholdings.com>\n> **Subject:** Re: Pioneer Building Open AI lease comments\n> **Date:** Monday, June 13, 2016 12:06 PM\n>\n> Happy to do either. The check would arrive tomorrow, the wire today. Your call.\n\n> **From:** Atit Jariwala <ajariwala@bridgetonholdings.com>\n> **To:** Jared Birchall <jbirchall@muskfoundation.org>\n> **Subject:** Re: Pioneer Building Open AI lease comments\n> **Date:** Monday, June 13, 2016\n>\n> Typically we prefer wires for payments, but in the case of first month's rent, we typically get a check to show our lender. That said I don't think it really matters. If it's easier for you to send a wire vs an overnight check, then that should be fine. Please let me know.\n\n> **From:** Jared Birchall <jbirchall@muskfoundation.org>\n> **To:** Atit Jariwala <ajariwala@bridgetonholdings.com>\n> **Subject:** Re: Pioneer Building Open AI lease comments\n> **Date:** Monday, June 13, 2016 3:03 AM\n>\n> Do you specifically want a check for the first months rent? Or would a wire work as well?\n\n> **From:** Jared Birchall <jbirchall@muskfoundation.org>\n> **To:** Atit Jariwala <ajariwala@bridgetonholdings.com>\n> **Subject:** Re: Pioneer Building Open AI lease comments\n> **Date:** Saturday, June 11, 2016\n>\n> Working on it, thanks! Quick question, I received an email from someone with OpenAI who was concerned because one of the Stripe people told him Charlie Weathers (CBRE broker) had shared that the new lease had been granted to Checkr. I'm assuming that someone is confused, but just wanted to get clarification directly from you.\n>\n> Thanks,\n> Jared\n\n> **From:** Atit Jariwala <ajariwala@bridgetonholdings.com>\n> **To:** Jared Birchall <jbirchall@muskfoundation.org>\n> **Subject:** Re: Pioneer Building Open AI lease comments\n> **Date:** Friday, June 10, 2016\n>\n> Jared, thanks – I can confirm receipt of the lease and guarantee. Yes, please do track those down as we'll need that to finalize this process.\n>\n> The below is correct – check made to Bridgeton Pioneer Property, LLC, which is the property owner.\n>\n> Our address is below. You can send to my attention. Thanks and have a great weekend!\n>\n> Atit M. Jariwala\n> Bridgeton Holdings LLC\n> 220 5th Avenue, 19th Floor\n> New York, NY 10001\n\n> **From:** Jared Birchall <jbirchall@muskfoundation.org>\n> **To:** Atit Jariwala <ajariwala@bridgetonholdings.com>\n> **Subject:** Re: Pioneer Building Open AI lease comments\n> **Date:** Friday, June 10, 2016 6:28 PM\n>\n> I assume the check is to be made out to Bridgeton Pioneer Property, LLC?\n>\n> And what is the address where I can send it?\n>\n> Thanks!\n\n> **From:** Atit Jariwala <ajariwala@bridgetonholdings.com>\n> **To:** Jared Birchall <jbirchall@muskfoundation.org>\n> **Subject:** Re: Pioneer Building Open AI lease comments\n> **Date:** Fri, Jun 10, 2016 at 12:05 PM\n>\n> Confirmed - that's correct on the first month's rent. Thanks\n\n> **From:** Jared Birchall <jbirchall@muskfoundation.org>\n> **To:** Atit Jariwala <ajariwala@bridgetonholdings.com>\n> **Subject:** Re: Pioneer Building Open AI lease comments\n> **Date:** Friday, June 10, 2016 3:04 PM\n>\n> Thanks! And just to confirm, the first month is $141,666.67, correct?\n\n> **From:** Atit Jariwala <ajariwala@bridgetonholdings.com>\n> **To:** Jared Birchall <jbirchall@muskfoundation.org>\n> **Subject:** Re: Pioneer Building Open AI lease comments\n> **Date:** Fri, Jun 10, 2016 at 11:41 AM\n>\n> Attached are the revised Lease and Guarantee removing Sam. The only things required to complete are: (1) signatures on both documents, (2) $400k LC or deposit check/wire, and (3) first month's rent check.\n>\n> Thanks, Jared,\n> Atit\n\n> **From:** Jared Birchall <jbirchall@muskfoundation.org>\n> **To:** Atit Jariwala <ajariwala@bridgetonholdings.com>\n> **Subject:** Re: Pioneer Building Open AI lease comments\n> **Date:** Friday, June 10, 2016 1:43 PM\n>\n> I spoke to Elon about it and he has asked that only he be on the lease and the guarantor.\n\n> **From:** Atit Jariwala <ajariwala@bridgetonholdings.com>\n> **To:** Jared Birchall <jbirchall@muskfoundation.org>\n> **Subject:** Re: Pioneer Building Open AI lease comments\n> **Date:** Friday, June 10, 2016\n>\n> Thanks for sharing those financials. They should work fine. If the lender has any issues, I'll let you know but it should be fine.\n>\n> Sam is actually not in the lease, just the guarantee so I'm hoping that's okay – Sam is the one that told me that he and Elon would both be on the guarantee. We would prefer to have him on the guarantee as well given his access to YC companies that may need extra space, if that need should arise in the future.\n>\n> Attached are final PDF's as well as a redline from the last version showing the small edits and inclusion of the Letter of Credit option in lieu of cash deposit.\n>\n> Looking forward to wrapping this up; please let me know if you have any questions,\n>\n> Atit\n\n> **From:** Jared Birchall <jbirchall@muskfoundation.org>\n> **To:** Atit Jariwala <ajariwala@bridgetonholdings.com>\n> **Subject:** Re: Pioneer Building Open AI lease comments\n> **Date:** Friday, June 10, 2016 3:07 AM\n>\n> **Also, Elon said he only wants the Musk Foundation on the Lease. Any problem removing Sam's name?**\n>\n> Thanks,\n> Jared\n\n> **From:** Jared Birchall <jbirchall@muskfoundation.org>\n> **To:** Atit Jariwala <ajariwala@bridgetonholdings.com>\n> **Subject:** Re: Pioneer Building Open AI lease comments\n> **Date:** Thursday, June 9, 2016\n>\n> Pretty simple, but this should give them what they need.\n>\n> I'm confirming the preference between cash vs letter of credit for the security deposit. For now you could put either/or on the contract. And we'll make good on it by tomorrow.\n>\n> Please let me know if you need anything else.\n\n> **From:** Atit Jariwala <ajariwala@bridgetonholdings.com>\n> **To:** Jared Birchall <jbirchall@muskfoundation.org>\n> **Subject:** Re: Pioneer Building Open AI lease comments\n> **Date:** Thu, Jun 9, 2016 at 3:52 PM\n>\n> Hi Jared,\n>\n> In the financials statements or letter you provide if you could list major liabilities as well or note that there are none that would be great.\n>\n> Thanks\n>\n> Sent from my iPhone\n\n> **From:** Jared Birchall <jbirchall@muskfoundation.org>\n> **To:** Atit Jariwala <ajariwala@bridgetonholdings.com>\n> **Subject:** Re: Pioneer Building Open AI lease comments\n> **Date:** Thursday, June 9, 2016 4:33 PM\n>\n> The official names are as follows:\n>\n> Musk Foundation\n> OpenAI, Inc.\n>\n> I somehow deleted your last email. Was this all you needed?\n\n> **From:** Atit Jariwala <ajariwala@bridgetonholdings.com>\n> **To:** Jared Birchall <jbirchall@muskfoundation.org>\n> **Subject:** Re: Pioneer Building Open AI lease comments\n> **Date:** Wed, Jun 8, 2016 at 6:09 PM\n>\n> Hi Jared,\n>\n> Please see attached. I made the adjustments discussed below and as per our call. Please let me know if you have any comments to this or if you think I missed something.\n>\n> On the parking, I'm still finding out more. But there is a public parking garage on Mission Street between 19th and 20th Streets and some private lots nearby, as well as street parking. That said the vast majority of the current tenant's employees take the bus, which has a bus stop literally at the same corner as the building, take the BART, which is a 10-minute walk away, or bike. They actually keep a lot of bike storage on the lot. The current tenant actually installed a large tent over nearly the entire parking lot, so they don't really need any parking spots but rent the spots for their tent. And of course, many folks use Uber and other ride sharing services. As I find out more, I'll let you know.\n>\n> Please let me know if you have any questions,\n> Atit\n\n> **From:** Jared Birchall <jbirchall@muskfoundation.org>\n> **To:** Atit Jariwala <ajariwala@bridgetonholdings.com>\n> **Subject:** Re: Pioneer Building Open AI lease comments\n> **Date:** Wednesday, June 8, 2016 6:55 PM\n>\n> OK, great. Do you want me to modify the lease document per our discussed changes? Or did you want to do that? Once we can get those modifications in place we are ready to move forward.\n\n> **From:** Atit Jariwala <ajariwala@bridgetonholdings.com>\n> **To:** Jared Birchall <jbirchall@muskfoundation.org>\n> **Subject:** Re: Pioneer Building Open AI lease comments\n> **Date:** Wednesday, June 8, 2016\n>\n> I'll get back to you on where the current tenant parks – they have a lot of employees in multiple buildings nearby and this is the headquarters where everyone comes. I'll find out where they park and where others park in the neighborhood.\n>\n> You can probably stack the parking, but this is something we have not looked in to.\n>\n> Thanks,\n> Atit\n\n> **From:** Jared Birchall <jbirchall@muskfoundation.org>\n> **To:** Atit Jariwala <ajariwala@bridgetonholdings.com>\n> **Subject:** Re: Pioneer Building Open AI lease comments\n> **Date:** Wednesday, June 08, 2016 6:03 PM\n>\n> Ok. Elon was a bit concerned when he heard there were only 23 spaces. Any information you have regarding alternatives in the near vicinity would be very helpful.\n>\n> Thank you,\n> Jared\n\n> **From:** Atit Jariwala <ajariwala@bridgetonholdings.com>\n> **To:** Jared Birchall <jbirchall@muskfoundation.org>\n> **Subject:** Re: Pioneer Building Open AI lease comments\n> **Date:** Wed, Jun 8, 2016 at 2:40 PM\n>\n> Hi Jared,\n>\n> The current tenant actually took 2 of the spaces to make the outdoor patio. That is why there are fewer open spaces for parking. The patio can be removed for the additional parking spaces, if needed.\n>\n> Thanks,\n> Atit\n\n> **From:** Jared Birchall <jbirchall@muskfoundation.org>\n> **To:** Atit Jariwala <ajariwala@bridgetonholdings.com>\n> **Subject:** Re: Pioneer Building Open AI lease comments\n> **Date:** Wednesday, June 08, 2016 5:19 PM\n>\n> Quick question, the website says there are 25 parking stalls. Any reason why only 23 are being offer in the lease agreement?\n>\n> Thanks!\n> Jared\n\n> **From:** Atit Jariwala <ajariwala@bridgetonholdings.com>\n> **To:** Jared Birchall <jbirchall@muskfoundation.org>\n> **Subject:** Re: Pioneer Building\n> **Date:** Wed, Jun 8, 2016 at 8:55 AM\n>\n> Hi Jared,\n>\n> I look forward to our call in approximately 1.5 hours. In the meanwhile, I wanted to send you responses to your below notes, which we can discuss on the call. I actually did speak with Elon about the additional rent but at a high level – only that there is additional items but it's not very material. The additional rent in year 1 likely will add up to $1,500 or so per month. These include items like increases in costs of repairs, security, common area utilities, insurance, trash removal, etc. Again, these are increases from prior year's expenses so the increase in cost is usually very nominal from year to year – typically the cost of inflation.\n>\n> I look forward to our conversation.\n>\n> All the best,\n> Atit\n\n> **From:** Jared Birchall <jbirchall@muskfoundation.org>\n> **To:** Atit Jariwala <ajariwala@bridgetonholdings.com>\n> **Subject:** Re: Pioneer Building\n> **Date:** June 7, 2016 at 2:34:59 PM EDT\n>\n> Hello Atit - Below is a list of my concerns/questions with regard to the lease. It would be great to speak asap. Please let me know what time can work for you?\n>\n> Thank you,\n> Jared\n>\n> 1. Already mentioned in my prior email, but the 1st full sentence on the top of page ii reads: \"The Base Rent for the renewal period shall be….\" If you'd like to discuss terms for renewal beyond the initial 10 years, happy to do it. But the terms listed (greater of 100% of market value or 3% higher than last year lease) **aren't terms we can accept. We will remove this from the lease. The full term will be 10 years and if we want to discuss a renewal in the future we can paper that with an amendment.**\n>\n> 2. Also already mentioned, there are numerous mentions of \"Additional Rent\" - which I'm assuming **Elon wasn't accounting for when he was negotiating the lease rate with you.** You said you would get back to me with an estimate on what this may amount to. Each of the following points relate to this issue.\n>\n> - The \"Additional Rent\" charge for the 23 parking spaces is not disclosed. The charge for the parking spaces should be disclosed and be limited at most to the fair rental value for the spaces. *(Atit reply: The charge for the parking would be $275/month adjusted to fair market rent on an annual basis. We are currently getting $275/month, however market is closer to $325/month.)*\n>\n> - Additional Rent includes all increases in real property taxes over the 2016 base period taxes. If a change in ownership occurs after the Lease commences, then we would pay the entire property tax increase over the base period. That tax increase could be very significant. *(Atit reply: This is San Francisco market standard for office leases so we are just requesting for what's market.)*\n>\n> - It's not clear if the Tenant has to pay the Landlord's liability for the San Francisco Gross Receipts Tax on rents. The tax rate is only 0.285% of rent received, so it's not significantly worrisome - just important to clarify. *(Atit reply: Tenant does not have to pay the Gross Receipts Tax on rents.)*\n>\n> - Tenant has to pay for some capital costs of the building. Section 4(b) says that the Tenant has to pay the \"replacement costs\" of all the Building's Systems defined in Section 1, such as, HVAC, elevators, plumbing and sprinklers. That replacement cost ought to be limited to a reasonable amortized cost assigned only to the remaining years on the Lease. For example, what if all the elevators or the HVAC system are replaced in the last year of the lease. *(Atit reply: Agree. We will make the change and Tenant will only be responsible for the amortized cost if replacement is not due to Tenant's fault.)*\n>\n> - The capital improvement costs at the end of the first sentence in Section 4(c)(i),(ii) and(iii) are amortized over the reasonable life of the improvement. The Building Systems replacement costs ought to be limited in the same way. *(Atit reply: Agree. We will make the change.)*\n>\n> - The capital costs listed at the first sentence in Section 4(c)(i),(ii) and(iii) are usually included in current office leases and themselves are ok. *(Atit reply: Great.)*\n>\n> 3. Tenant pays for all insurance, including earthquake insurance if Landlords chooses to get it. Earthquake insurance can be very expensive and, in general, **Elon hates insurance. I am doubtful that Elon will agree to this provision.** *(Atit reply: Similar to property taxes, this is also a market standard provision. We can discuss this and maybe its balanced out somewhere else.)*\n\n> **From:** Atit Jariwala <ajariwala@bridgetonholdings.com>\n> **To:** Jared Birchall <jbirchall@muskfoundation.org>\n> **Subject:** Re: Pioneer Building\n> **Date:** Tue, Jun 7, 2016 at 9:59 AM\n>\n> Jared,\n>\n> Please let me know if you have any comments. I'm running out of the office now for a conference so will be difficult to reach today via phone. I'm around tomorrow, though, should you want to speak.\n>\n> All the best,\n> Atit\n\n> **From:** Jared Birchall <jbirchall@muskfoundation.org>\n> **To:** Atit Jariwala <ajariwala@bridgetonholdings.com>\n> **Subject:** Re: Pioneer Building\n> **Date:** Sunday, June 05, 2016 7:40 PM\n>\n> Atit - Actually, two questions in advance of speaking:\n>\n> - Per the contract, the 'base rent' is represented by what you and Elon agreed to over email. Do you have a broad estimate on what the 'Additional Rent' might amount to on a monthly basis. I understand that there are some variables with the insurance we choose and the potential maintenance needs, but the taxes should be known. I'm just trying to get a ballpark idea.\n>\n> - Renewal of the lease - Do I understand correctly where on the top of page 3 the contract attempts to set the terms for renewing the lease in the future? The renewal would be granted at the greater of (i) 100% of the current market value or (ii) a 3% increase over your final year's lease payment?\n>\n> Thank you,\n>\n> Jared\n\n> **From:** Atit Jariwala <ajariwala@bridgetonholdings.com>\n> **To:** Jared Birchall <jbirchall@muskfoundation.org>\n> **Subject:** Re: Pioneer Building\n> **Date:** Sun, Jun 5, 2016 at 4:34 PM\n>\n> Hi Jared,\n>\n> Nice to meet you. I look forward to working with you and wrapping this up. I am available should you have any questions - my cell is 646.554.5400 and my office line is 212.235.2780. Please don't hesitate to reach out.\n>\n> All the best,\n> Atit\n>\n> Sent from my iPhone\n\n> **From:** Jared Birchall <jbirchall@muskfoundation.org>\n> **To:** Atit Jariwala <ajariwala@bridgetonholdings.com>\n> **Subject:** Re: Pioneer Building\n> **Date:** Sun, Jun 5, 2016 at 6:33 PM\n>\n> Hello Atit - I hope you are well. I am in the process of reviewing the lease agreement and will get back to you by tomorrow. Our intention is clearly to make this happen. If I have any questions, will you be available to speak sometime tomorrow?\n>\n> Thank you,\n> Jared\n\n> **From:** Elon Musk <elon@openai.com>\n> **To:** Jared Birchall <jbirchall@muskfoundation.org>\n> **Subject:** Fwd: Pioneer Building\n> **Date:** Fri, Jun 3, 2016 at 12:51 PM\n>\n> Please review\n\n> **From:** Atit Jariwala <ajariwala@bridgetonholdings.com>\n> **To:** Elon Musk <elon@openai.com>\n> **Subject:** RE: Pioneer Building\n> **Date:** June 3, 2016 at 4:41:38 AM PDT\n>\n> Good morning – see attached lease draft. Please let me know if you have any comments/questions.\n\n## Commentary\n\nPX 75 is the documentary spine of plaintiffs' Pioneer Building chain — the 2016 origination of the lease that defense will later say (per [[DX-539]]) was a Musk-Foundation-owned arrangement, not an OpenAI charitable asset. The exhibit cuts both ways. **Plaintiffs' read:** the building was negotiated by the Musk Foundation as a building \"to house OpenAI, Neuralink and maybe some SpaceX or Tesla people,\" and Musk personally directed Birchall on June 10, 2016 to **remove Sam Altman's name** (\"Elon said he only wants the Musk Foundation on the Lease. Any problem removing Sam's name?\") — a control marker that becomes important when the rent reimbursement question arises in 2017–2020 (see [[PX-103]] for the July 2020 Clark email and the § 17510.6 charitable-solicitation theory). **Defense read:** the email **\"this should be viewed as a Musk Foundation building\"** ([[DX-539]]) is corroborated here by the wiring chain — the first month's $141,666.67 went to Bridgeton Pioneer Property, LLC from a Musk-side account, and Birchall executed the entire negotiation from `jbirchall@muskfoundation.org` with no OpenAI signatory, confirming the building is Musk-side multi-tenant infrastructure rather than an OpenAI nonprofit asset. The `elon@openai.com` June 3 forwarding email is also notable — it places Musk personally negotiating directly with the landlord using his OpenAI address, a fact plaintiffs use to argue OpenAI was the real beneficiary.\n\n---\n*See also:* [[Day 4|Day 4 digest]] · [[Pioneer Building]] · [[Jared Birchall]] · [[DX-539]] · [[PX-103]] · [[Statute of Limitations]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-79", "exhibit": "PX 79", "party": "Plaintiffs", "type": "Lease", "admitted_trial_day": "Day 4 (April 30, 2026)", "uploaded_box_pt": "2026-04-30T15:26:52", "uploader": "Someone", "pages": 51, "size_bytes": 1163372, "source_pdf": "PX-79.pdf", "pdf_url": "https://media.mts-in.com/PX-79.pdf", "body_markdown": "# PX 79 — Pioneer Building master lease: Bridgeton Pioneer Property LLC ↔ Musk Industries LLC (June 17, 2016), with Elon Musk personal guaranty\n\n> The fully executed master office lease for **The Pioneer Building, 3180 18th Street, San Francisco** — the building OpenAI occupied — with **Musk Industries LLC** (not OpenAI) as tenant and **Elon Musk** as personal guarantor; 10-year term, 37,104 RSF, escalating from $850K to $3.42M annual base rent.\n\n## Document type\n**Lease** (Office Lease Agreement) plus a Birchall → Musk transmittal email (Bridgeton's Atit Jariwala forwarding the fully-executed lease for \"Lender Approval\"), the lease itself with full Basic Lease Information and 25 sections, and the form of personal Guaranty (Exhibit G) signed by Elon Musk on June 17, 2016. Bates SPX-002475–002525.\n\n## Logistics\n- **Trial admission:** Day 4 (April 30, 2026) — used in Musk cross / Birchall direct in the [[Key Themes]] §\"Pioneer Building\" sequence to establish that Musk Industries LLC, not OpenAI, was the master lessee, and that Musk personally guaranteed the lease. Foundation for plaintiffs' alternative § 17510.6 charitable-solicitation theory.\n- **Box upload:** 2026-04-30 15:26:52 PT — Day 4 mid-afternoon Plaintiffs' batch (uploaded one second before [[PX 60]]).\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~1.1 MB, 51 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `079.pdf`.\n\n## Transcribed text\n\n> **Email — Jared Birchall <jbirchall@muskfoundation.org> → Elon Musk; Cc Sam Teller**\n> **Subject:** Pioneer Building Lender Approval\n> **Date:** Fri, 1 Jul 2016 12:28:34 -0700\n> **Attachments:** Pioneer_Open_AI_Lease_Fully_Executed.pdf\n>\n> Finally!! I just got off the phone with Atit and he's excited to move forward. He will facilitate a site inspection as soon as we'd like.\n>\n> ---------- Forwarded message ----------\n> From: **Atit Jariwala <ajariwala@bridgetonholdings.com>** — Friday, July 1, 2016 — Subject: Lender Approval — To: Jared Birchall:\n>\n> Hi Jared,\n> Attached is the fully executed lease. Congrats! We're excited to have you guys in the space. I typically like to setup closing dinners — maybe we can do something someday when we're in the same city. Let's talk about next steps. Thanks, Atit.\n>\n> -- *Jared J. Birchall, Managing Director, Musk Family Office*\n\n> **THE PIONEER BUILDING — 3180 18TH STREET — OFFICE LEASE AGREEMENT** between **BRIDGETON PIONEER PROPERTY LLC**, a Delaware LLC (\"Landlord\"), and **MUSK INDUSTRIES LLC**, a California LLC (\"Tenant\"). Dated as of **June 17, 2016**.\n>\n> **Basic Lease Information**:\n> - **Premises:** 37,104 rentable square feet at 3180 18th Street, San Francisco, CA 94110 (\"The Pioneer Building\").\n> - **Term:** 10 years, commencing August 15, 2016; expiring 5:00 p.m. on the last day of the 120th full calendar month.\n> - **Base Rent schedule:**\n>   - Months 1–6: $141,666.67/mo ($850,000)\n>   - Months 7–12: $175,000.00/mo ($1,050,000)\n>   - Months 13–18: $208,333.33/mo ($1,250,000)\n>   - Months 19–24: $241,666.67/mo ($1,450,000)\n>   - Months 25–36: $231,750.00/mo ($2,781,000)\n>   - Months 37–48: $238,702.50/mo ($2,864,430)\n>   - Months 49–60: $245,863.58/mo ($2,950,362.90)\n>   - Months 61–72: $253,239.48/mo ($3,038,873.79)\n>   - Months 73–84: $260,836.67/mo ($3,130,040.00)\n>   - Months 85–96: $268,661.77/mo ($3,223,941.20)\n>   - Months 97–108: $276,721.62/mo ($3,320,659.44)\n>   - Months 109–120: $285,023.27/mo ($3,420,279.22)\n> - **Security Deposit:** $400,000 (cash or letter of credit).\n> - **Guarantor:** Elon Musk.\n> - **Permitted Use:** General office use.\n> - **Parking:** Up to 23 spaces, $275/space/month.\n> - **Tenant's address (pre-Commencement):** c/o Musk Industries LLC, 1 Rocket Road, Hawthorne, CA 90250, Attn: Elon Musk. (Post-Commencement: 3180 18th Street, Suite 1, San Francisco.)\n\n> **Section 25(y) — Tenant Guaranty; OpenAI, Inc.** (Lease pp. 24–25):\n> \"Tenant and Elon Musk hereby represent that Tenant is **(i) wholly-owned by Elon Musk** and (ii) in good standing in the State of California. By signing below, Elon Musk agrees and covenants to Landlord that (i) **he shall maintain sole ownership of Tenant at all times during the term of the Lease**, (ii) he absolutely, unconditionally and irrevocably guarantees to Landlord, and agrees fully to pay, perform, satisfy and discharge … all of the indebtedness, agreements, obligations and liabilities of Tenant under this Lease… (and Landlord may proceed directly against Elon Musk with respect thereto)…\n>\n> At any time during the term of the Lease, **Tenant may allow OpenAI, Inc., a California nonprofit public benefit corporation, to use all or any part of the Premises for general office use; provided, however, at any time that OpenAI, Inc. uses the Premises, Tenant shall remain responsible for compliance with this Lease in all respects.** For any period in which OpenAI, Inc. uses and occupies the Premises, Landlord shall accept payment of Rent and other charges set forth in Section 4 of the Lease directly from OpenAI, Inc., and all other costs and charges payable by Tenant under the Lease; provided, however, **Tenant shall be solely responsible for such obligations under all circumstances**.\"\n\n> **Signatures** (June 17, 2016 / July 1, 2016): Atit Jariwala, Authorized Signatory, Bridgeton Pioneer Property LLC; Elon Musk, on behalf of Musk Industries LLC; and Elon Musk, \"Acknowledged and Agreed solely for purposes of Section 25(y).\"\n\n> **Exhibit G — Form of Guaranty (Guaranty of Lease)**:\n> \"For valuable consideration, … and to induce Landlord to enter into that certain Lease Agreement dated as of June __, 2016 (the 'Lease') herewith with **Musk Industries LLC** (the 'Tenant'), **Guarantor [Elon Musk] hereby absolutely, unconditionally and irrevocably guarantees** to Landlord, and agrees fully to pay, perform, satisfy and discharge, as and when payment, performance, satisfaction and discharge are due, all of the indebtedness, agreements, obligations and liabilities of the Tenant under the Lease… The obligations of Guarantor under this Guaranty shall be absolute, unconditional and irrevocable…\"\n>\n> Signature line: ELON MUSK.\n\n> [Remainder is the standard body of a commercial office lease — Sections 1–24 (definitions, AS-IS condition, services and utilities, alterations and signs, assignment and subletting (no consent for Affiliates), insurance, subordination, condemnation, casualty, defaults and remedies, holding over, surrender, hazardous materials, parking) and Exhibits A (Diagram of Premises — first floor plan), B (Standards for Services and Utilities), C (Building Rules and Regulations — including a 60-pound dog policy), D (Parking Rules and Regulations), E (Form of Commencement Date Memorandum), F (Form of Tenant Estoppel Certificate), G (the Guaranty above).]\n\n## Commentary\n\nPX 79 is the document plaintiffs use to flip the \"Musk gave OpenAI a building\" narrative on its head: the lessee is **Musk Industries LLC**, a passthrough Musk owned 100% (the \"wholly-owned by Elon Musk\" rep in §25(y) is admitted on the face of the lease), and the personal guarantor is Musk himself — OpenAI was permitted only to \"use\" the premises and to remit rent to Landlord, with Tenant remaining \"solely responsible … under all circumstances.\" Combined with [[DX 539]] (\"**I don't want Sam on the lease**\", June 2016) and [[DX 600]] (\"Elon holds **100 percent of the power and authority** regarding the building,\" Birchall to Sam Teller, January 2017), the lease cements defense's response on cross: Musk treated the Pioneer Building as his asset, not OpenAI's, even while the foundation paid the rent through donor-advised-fund grants documented in [[PX 60]]. The recurring **$175,000/month** Vanguard grants in 2017 line up almost exactly to the months 7–12 base-rent column ($175,000/mo). For the §17510.6 angle, see [[Key Themes]] §\"Pioneer Building\" — particularly **PX 103** (July 2020 Clark email asserting the for-profit OpenAI was \"happy to cover 100 percent of the rent\"), the alleged 2020 charitable-solicitation moment.\n\n---\n*See also:* [[Day 4|Day 4 digest]] · [[PX 60]] · [[PX 103]] · [[DX 539]] · [[DX 600]] · [[Jared Birchall]] · [[Elon Musk]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-8", "exhibit": "PX 8", "party": "Plaintiffs", "type": "Email thread", "admitted_trial_day": "Day 6 (May 4, 2026)", "uploaded_box_pt": "2026-05-04T14:46:03", "uploader": "Someone", "pages": 2, "size_bytes": 636830, "source_pdf": "PX-8.pdf", "pdf_url": "https://media.mts-in.com/PX-8.pdf", "body_markdown": "# PX 8 — Oct 26, 2015 Brockman → Marissa Mayer \"yc ai ask\"\n\n> Brockman's October 2015 fundraising email to Marissa Mayer soliciting OpenAI's first wave of $100M-targeted donations — the email in which he writes that \"Sam's donating $10M, I'm personally doing donating $100K\" and lists \"Elon Musk, Jessica Livingston, Reid Hoffman, and Peter Thiel\" among current donors.\n\n## Document type\n**Email thread, two messages.** Brockman's reply on top of his original Oct 20, 2015 8:55 AM solicitation to Nimit Jain (BCC: Sam Altman). Brockman is \"former CTO of Stripe… now running the first group in YC Research… a non-profit AI lab building towards artificial general intelligence.\" Bates OPENAI_MUSK00021496–97.\n\n## Logistics\n- **Trial admission:** Day 6 (May 4, 2026) — used during Mr. Kry's cross of Brockman to walk through the Oct 2015 solicitation in which Brockman represented he was \"personally donating $100K.\" Brockman conceded on the stand: \"Did not end up donating. That's true.\" Q: \"So that was a false statement.\" — A: \"I disagree with that characterization.\" (See [[Key Themes]] §\"Greg Brockman's contributions ('zero')\".)\n- **Box upload:** 2026-05-04 14:46:03 PT — same-day Day-6 morning batch alongside other Brockman-cross materials.\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~622 KB, 2 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `008.pdf`.\n\n## Transcribed text\n\n> **From:** Greg Brockman [redacted]\n> **on behalf of:** Greg Brockman [redacted]\n> **Sent:** 10/26/2015 2:56:27 AM\n> **To:** Nimit Jain [redacted]\n> **BCC:** Sam Altman [sam@ycombinator.com]\n> **Subject:** Re: yc ai ask\n>\n> > Hey, just have a few follow-on questions to bring some clarity to the ask:\n> >\n> > How much has each donor contributed?\n>\n> We're targeting a total of $100M — out of respect for the individuals, the breakdown is private.\n>\n> > How many funders do you foresee having ultimately?\n>\n> We're talking to handful of other folks (fewer than 5 in flight right now); in the interest of simplicity I don't expect to grow much beyond that.\n>\n> > What is your anticipated engagement model with donors after the initial contributions?\n>\n> We understand our donors are all super busy, and we don't expect time from them.\n>\n> > Who are the current candidates for lead scientist?\n>\n> We have an extremely well-respected person lined up as lead scientist. He's currently extricating himself from his employer, and being very careful about how the news breaks. I should be able to share his name within the next few weeks.\n>\n> Let me know if I can clarify anything further!\n>\n> - gdb\n>\n> ---\n>\n> > On Tue, Oct 20, 2015 at 8:55 AM, Greg Brockman <[redacted]> wrote:\n> >>\n> >> Hi Marissa,\n> >>\n> >> I'm the former CTO of Stripe, and I'm now running the first group in YC Research (http://blog.ycombinator.com/yc-research), a non-profit AI lab building towards artificial general intelligence.\n> >>\n> >> On the funding side, we have a small group of really fantastic individuals donating, including **Elon Musk, Jessica Livingston, Reid Hoffman, and Peter Thiel.** Because AI is something that should be built on behalf of all humans rather than any particular group, we're looking to broaden our set of funders within the industry.\n> >>\n> >> We'd particularly love to have you involved. (We're open to pretty much any amount you feel comfortable with: **Sam's donating $10M, I'm personally doing donating $100K** — what would be most impactful would be having you officially on as a donor.)\n> >>\n> >> For some more details, here's a rough draft of our launch blog post: https://ycr.quip.com/TQxVAXbubFqN. We've been quietly assembling a team of some of the best AI researchers and engineers in the world, and expect to proceed with transparency and that public announcement in the next month or two.\n> >>\n> >> Incidentally, the HN discussion of the YC Research announcement [2] was HN's all-time most highly-upvoted launch story. I expect the AI group in particular is going to get a lot of attention, and we're being very careful to get the best people involved and position it for maximal impact.\n> >>\n> >> I'm happy to chat at any time: [Redacted – PII] / gdb@ycr.org. In any case, thanks for the consideration!\n> >>\n> >> - gdb\n> >>\n> >> [1] https://ycr.quip.com/TQxVAXbubFqN\n> >> [2] https://news.ycombinator.com/item?id=10347821\n\n## Commentary\n\nPX 8 is the **founding solicitation document** of OpenAI's pre-launch fundraising — and one of plaintiffs' cleanest exhibits on Brockman's personal candor with donors. Three lines do the work: (i) \"we're targeting a total of $100M\" — the original $100M aspiration that becomes the public $1B Musk pledge two months later; (ii) \"Sam's donating $10M, I'm personally doing donating $100K\" — Brockman's representation of his own contribution, which on Day 6 he conceded he never paid; (iii) \"Elon Musk… [is] donating\" — the email is sent to Mayer using Musk's name as a credentialing fact while Musk is unaware of the solicitation's specifics. The exhibit is the documentary anchor for plaintiffs' \"Greg Brockman's contributions = zero\" theme (see [[Key Themes]]) and slots between [[Greg Brockman]]'s Stripe-CTO bona fides and the December 2015 OpenAI launch. Cross-reference [[Brockman Journal]] for Brockman's contemporaneous internal writing in the same window.\n\n---\n*See also:* [[Day 6|Day 6 digest]] · [[Greg Brockman]] · [[Sam Altman]] · [[Brockman Journal]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-80", "exhibit": "PX 80", "party": "Plaintiffs", "type": "Email thread", "admitted_trial_day": "Day 4 (April 30, 2026)", "uploaded_box_pt": "2026-04-30T15:26:56", "uploader": "Someone", "pages": 3, "size_bytes": 232718, "source_pdf": "PX-80.pdf", "pdf_url": "https://media.mts-in.com/PX-80.pdf", "body_markdown": "# PX 80 — July 9–10, 2016 Musk ↔ Birchall \"Re: Weekly Update\" (DAF / 501(c)(3) conduit)\n\n> A three-page weekly-update email thread between Elon Musk and Jared Birchall in which Birchall explains that OpenAI's quarterly contributions from Musk were being routed through the Musk Foundation DAF \"channeled through the temporary 501(c)(3) that they are using\" because OpenAI did not yet have its own 501(c)(3) status — and that \"**The plan is to always give through the DAF so no press is generated on that front**.\"\n\n## Document type\n**Email thread, plain text, four-message chain.** Musk's \"Good\" reply on top of three earlier Birchall ↔ Musk weekly-update messages from July 9–10, 2016 — including a Birchall balance-sheet email with an attached `Balance sheet EM 7.8.xlsx` file. Subject \"Re: Weekly Update.\" Bates 2024MUSK-0009632–9634. Portions are redacted in black-box overlays.\n\n## Logistics\n- **Trial admission:** Day 4 (April 30, 2026 — Musk cross + redirect; Birchall direct + cross). Per the Day 4 transcript table: \"PX 80 — Birchall email noting YC Org used as 'another 501(c)(3) as a conduit'\" `(043026TT.txt:4475–4486)`.\n- **Box upload:** 2026-04-30 15:26:56 PT — Day 4 late-afternoon plaintiffs' batch.\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~227 KB, 3 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `080.pdf`.\n\n## Transcribed text\n\n> **From:** Elon Musk <erm@spacex.com>\n> **To:** Jared Birchall <jbirchall@muskfoundation.org>\n> **Subject:** Re: Weekly Update\n> **Date:** Sun, 10 Jul 2016 16:02:48 -0000\n>\n> Good\n\n> **From:** Jared Birchall <jbirchall@muskfoundation.org>\n> **To:** Elon Musk <erm@spacex.com>\n> **Subject:** Re: Weekly Update\n> **Date:** Sun, Jul 10, 2016 at 5:43 AM\n>\n> OpenAI has received $5.5MM from you. But it was sent from the DAF and was channeled through the temporary 501(c)(3) that they are using. Anyone who has given any money to OpenAI has gone the same route at the is point.\n\n> **From:** Elon Musk <erm@spacex.com>\n> **To:** Jared Birchall <jbirchall@muskfoundation.org>\n> **Subject:** Re: Weekly Update\n> **Date:** Sunday, July 10, 2016\n>\n> So I haven't sent anything to OpenAI?\n>\n> That's a really big deal. My credibility is at stake here.\n\n> **From:** Jared Birchall <jbirchall@muskfoundation.org>\n> **To:** Elon Musk <erm@spacex.com>\n> **Subject:** Re: Weekly Update\n> **Date:** Sat, Jul 9, 2016 at 9:03 PM\n>\n> We were supposed to begin the quarterly contributions to OpenAI in April, but because they didn't have an entity in place to even make a contribution we didn't pay. They scrambled and in June started using another 501(c)(3) as a conduit until OpenAI is officially granted 501(c)(3) status. We verified the conduit and just made the first payment in June. The application process is long and tedious. I'm not sure why they have taken so long to apply, but my inquiry lit a fire under them. Earlier today I received all the electronic formation documents for OpenAI, as well as for the conduit organization and have started going through them. I will request the other docs as well.\n>\n> **The plan is to always give through the DAF so no press is generated on that front.** I'll keep tabs on other items that could potentially generate press.\n>\n> The building rent starts August 15th. Atit already has our first month's rent. He'll make the building available as needed for planning and prep.\n>\n> [black-box redaction]\n\n> **From:** Elon Musk <erm@spacex.com>\n> **To:** Jared Birchall <jbirchall@muskfoundation.org>\n> **Subject:** Re: Weekly Update\n> **Date:** Saturday, July 9, 2016\n>\n> Did we pay the last OpenAI installment at beginning of quarter? If so, we should continue on that track. I committed to $5M per quarter. It's fine to take the actions necessary for that support. Just need to know in advance if anything is going to generate press.\n>\n> Please ask for (ideally electronic) copies of everything at OpenAI, including all offer letters plus anything else and lmk if there is anything concerning. Who is responsible for administrative work there? I have had very little bandwidth to think about the company and am a little worried that it is being managed as an extension of YCombinator. The YC stock as comp for many who joined obviously incents them in that direction. Does everyone have YC stock? That would be weird.\n>\n> When exactly does the building become available and the rent start? I would like to use part of the building for Neuralink at some point, so no YC stuff.\n>\n> [black-box redaction]\n\n> **From:** Jared Birchall <jbirchall@muskfoundation.org>\n> **To:** Elon Musk <erm@spacex.com>\n> **Subject:** Weekly Update\n> **Date:** Sat, Jul 9, 2016 at 12:03 AM\n> **Attachments:** Balance sheet EM 7.8.xlsx\n>\n> Elon - I hope the trip has gone well! Attached is your weekly balance sheet update. You can 'ignore links' when it asks.\n>\n> **OpenAI -**\n> - I spoke with Chris Clark yesterday and he requested that we send the quarterly $5MM commitment at the beginning of the quarter - which is now. The only problem is we don't have sufficient cash to make that happen. We have $150k in the Foundation and $1.8 mil in the DAF. The foundation would need to sell some shares in order to generate the necessary cash. I share your desire to hold off for better valuations, but if we want to accommodate this request we'd need to gift 23k shares to the DAF where they would provide the payment. Thoughts?\n> - We will also have the ~$150k payable each month for rent. That's another $600k this year, in addition to the $600k we'll need to support the USC project (among a few other commitments). Plus another $10MM committed in January.\n> - Chris asked me about using the extra space in the building for some of the Y Combinator companies. I told him to hold off on any of those plans until I talk it through with you.\n> - As suggested, I have requested a copy of all the OpenAI formation documents and am awaiting these from Chris.\n>\n> [black-box redaction]\n>\n> Jared J. Birchall\n> Managing Director\n> **Musk Family Office**\n\n## Commentary\n\nPX 80 is plaintiffs' contemporaneous receipt that **Musk's OpenAI funding was DAF-laundered through a \"temporary\" third-party 501(c)(3) conduit** before OpenAI obtained its own tax-exempt status, and that the routing was deliberate — \"**The plan is to always give through the DAF so no press is generated on that front**.\" Plaintiffs use this for two purposes: (i) to show Musk personally controlled the funding mechanism and the press footprint of his giving (relevant to the [[Key Themes|\"$1B / $38M gap\"]] and \"donor-advised funds / tax-deduction\" themes — see Birchall's concession on cross that \"for every single one of those donations … Mr. Musk was entitled to take a tax deduction\"), and (ii) the **Pioneer Building rent** sub-thread (\"$150k payable each month for rent. That's another $600k this year\") establishes the same Musk-Foundation-as-landlord posture defense relies on in [[DX-539]] for its § 17510.6 counter-narrative — but plaintiffs use the same facts to argue the building was an OpenAI-charitable-asset solicitation. The \"**My credibility is at stake here**\" line shows Musk treating the donations as personal-reputation events. See [[Day 4|Day 4 digest]] for the full Birchall direct/cross arc and the ~4475–4486 transcript pages.\n\n---\n*See also:* [[Day 4|Day 4 digest]] · [[Jared Birchall]] · [[Pioneer Building]] · [[DX-539]] · [[PX 103]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-84", "exhibit": "PX 84", "party": "Plaintiffs", "type": "Email thread", "admitted_trial_day": "Day 4 (April 30, 2026)", "uploaded_box_pt": "2026-04-30T15:27:04", "uploader": "Someone", "pages": 8, "size_bytes": 1085118, "source_pdf": "PX-84.pdf", "pdf_url": "https://media.mts-in.com/PX-84.pdf", "body_markdown": "# PX 84 — Oct/Nov 2016 Birchall–Musk Family Office DAF stock-transfer email thread for \"the grant to OpenAI\"\n\n> Reverse-chron internal Musk Family Office / CTC|myCFO / Morgan Stanley email thread arranging the late-2016 transfer of ~26,000 shares (~$5.2M) from the Musk Foundation to a donor-advised fund to fund \"the grant to OpenAI.\"\n\n## Document type\n**Email thread, internal financial-services correspondence.** Multi-message reverse-chronological chain among Jared Birchall (Musk Family Office Managing Director, jbirchall@muskfoundation.org), Ronald Gong, Paula Lo and Teresa Holland (CTC | myCFO, BMO Financial Group), and Matilda Simon-Ferrigno (Morgan Stanley Private Wealth Management, Malone Neuhaus Group). Bates EXMF-0003982 through EXMF-0003989. Marked \"Highly Confidential-AEO.\"\n\n## Logistics\n- **Trial admission:** Day 4 (April 30, 2026) — Musk cross + redirect; Birchall direct + cross.\n- **Box upload:** 2026-04-30 15:27:04 PT.\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~1.06 MB, 8 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `084.pdf`.\n\n## Transcribed text\n\n> **From:** Jared Birchall <jbirchall@muskfoundation.org>\n> **To:** Paula Lo <[redacted]@ctcmycfo.com>\n> **Cc:** Ronald Gong <[redacted]@ctcmycfo.com>, Teresa Holland <[redacted]@ctcmycfo.com>, Harris myCFO MUS001FOS <MUS001FOS@ctcmycfo.com>\n> **Subject:** Re: projected giving - update\n> **Date:** 11/9/2016 11:28:50 PM\n>\n> Ok, thank you for the update, Paula!\n>\n> *Jared J. Birchall*\n> Managing Director\n> **Musk Family Office**\n\n> **From:** Paula Lo <[redacted]@ctcmycfo.com>\n> **To:** Jared Birchall <jbirchall@muskfoundation.org>\n> **Cc:** Ronald Gong <[redacted]@ctcmycfo.com>, Teresa Holland <[redacted]@ctcmycfo.com>, Harris myCFO MUS001FOS\n> **Subject:** Re: projected giving - update\n> **Date:** Wed, Nov 9, 2016 at 2:25 PM\n>\n> Hi Jared,\n>\n> We checked the account this morning and the additional ~$5M is not yet available for us to process the grant to Open AI.\n>\n> Vanguard received the shares on Tuesday. The shares are in the process of being sold and the trades are not settled yet.\n>\n> We are expecting the funds will be ready by early next week for us to process the grant payment.\n>\n> Best,\n> Paula\n>\n> Paula Lo | Director\n> CTC | myCFO | 2200 Geng Road, Suite 100 | Palo Alto, CA 94303\n> A Part of BMO Financial Group\n\n> **From:** Jared Birchall <jbirchall@muskfoundation.org>\n> **To:** Ronald Gong <[redacted]@ctcmycfo.com>\n> **Cc:** Teresa Holland <[redacted]@ctcmycfo.com>, Harris myCFO MUS001FOS\n> **Subject:** Re: projected giving - update\n> **Date:** Wednesday, November 09, 2016 1:55 PM\n>\n> Hi Ron - Any insight on the timing regarding sending the additional funds to OpenAI?\n>\n> Thanks,\n> Jared\n\n> **From:** Ronald Gong <[redacted]@ctcmycfo.com>\n> **To:** Matilda Simon-Ferrigno <[redacted]@morganstanleypwm.com>\n> **Cc:** Teresa Holland, Jared Birchall <jbirchall@muskfoundation.org>, Harris myCFO MUS001FOS\n> **Subject:** RE: projected giving - update\n> **Date:** Fri, Nov 4, 2016 at 3:17 PM\n>\n> Paula will send on Monday am. Thx\n>\n> Sent from my BlackBerry 10 smartphone.\n\n> **From:** Matilda Simon-Ferrigno <[redacted]@morganstanleypwm.com>\n> **To:** Ronald Gong <[redacted]@ctcmycfo.com>, Teresa Holland <[redacted]@ctcmycfo.com>\n> **Cc:** Jared Birchall <jbirchall@muskfoundation.org>, Harris myCFO MUS001FOS\n> **Subject:** RE: projected giving - update\n> **Date:** Friday, November 4, 2016 3:09 PM\n>\n> Hi Ron — yes, please send me the instructions to move the stock, thank you!\n>\n> Matilda Simon-Ferrigno\n> Group Director / Assistant Vice President\n> Morgan Stanley Private Wealth Management\n> 444 South Flower Street 34th Floor | Los Angeles, CA 90071\n> The Malone Neuhaus Group\n\n> **From:** Ronald Gong <[redacted]@ctcmycfo.com>\n> **To:** Matilda Simon-Ferrigno <[redacted]@morganstanleypwm.com>, Teresa Holland <[redacted]@ctcmycfo.com>\n> **Cc:** Jared Birchall <jbirchall@muskfoundation.org>, Harris myCFO MUS001FOS\n> **Subject:** Re: projected giving - update\n> **Date:** Friday, November 04, 2016 2:49 PM\n>\n> Hi Matilda - we would like to move these shares on Monday.\n>\n> Just let us know what you will need. You mention DAF paperwork, do you mean the stock transfer details?\n\n> **From:** Matilda Simon-Ferrigno <[redacted]@morganstanleypwm.com>\n> **To:** Ronald Gong <[redacted]@ctcmycfo.com>, Teresa Holland <[redacted]@ctcmycfo.com>\n> **Cc:** Jared Birchall <jbirchall@muskfoundation.org>\n> **Subject:** RE: projected giving - update\n> **Date:** Monday, October 31, 2016 10:42 AM\n>\n> Ok, great..!\n>\n> Matilda Simon-Ferrigno\n> Group Director / Assistant Vice President\n> Morgan Stanley Private Wealth Management\n\n> **From:** Ronald Gong <[redacted]@ctcmycfo.com>\n> **To:** Matilda Simon-Ferrigno <[redacted]@morganstanleypwm.com>, Teresa Holland <[redacted]@ctcmycfo.com>\n> **Cc:** Jared Birchall <jbirchall@muskfoundation.org>\n> **Subject:** RE: projected giving - update\n> **Date:** Monday, October 31, 2016 10:41 AM\n>\n> Hi Matilda — sounds good — probably toward the end of the week. Will let you know.\n>\n> thanks\n>\n> Ron Gong | Managing Director\n> CTC | myCFO | 2200 Geng Road, Suite 100 | Palo Alto, CA 94303\n\n> **From:** Matilda Simon-Ferrigno <[redacted]@morganstanleypwm.com>\n> **To:** Ronald Gong <[redacted]@ctcmycfo.com>, Teresa Holland <[redacted]@ctcmycfo.com>\n> **Cc:** Jared Birchall <jbirchall@muskfoundation.org>\n> **Subject:** RE: projected giving - update\n> **Date:** Monday, October 31, 2016 10:04 AM\n>\n> I can do this any time you are ready. I do not need to break the cert. Thx!\n>\n> Best,\n> Matilda\n\n> **From:** Matilda Simon-Ferrigno <[redacted]@morganstanleypwm.com>\n> **To:** Ronald Gong <[redacted]@ctcmycfo.com>, Teresa Holland <[redacted]@ctcmycfo.com>\n> **Cc:** Jared Birchall <jbirchall@muskfoundation.org>\n> **Subject:** FW: projected giving\n> **Date:** Monday, October 31, 2016 8:50 AM\n>\n> Hi Ron/Teresa — I hope you are both well. Please let me know how I can help. **I may need to break a certificate for 26k shs, which will take approximately 1 week.** Please forward the DAF paperwork when you can.\n>\n> Thanks!\n> Best,\n> Matilda\n\n> **From:** Ronald Gong <[redacted]@ctcmycfo.com>\n> **To:** Jared Birchall <jbirchall@muskfoundation.org>, Teresa Holland <[redacted]@ctcmycfo.com>\n> **Cc:** Paula Lo <[redacted]@ctcmycfo.com>\n> **Subject:** projected giving\n> **Date:** Friday, October 28, 2016\n>\n> Yes, that would be the amount. Can you please check w MS on my procedure to move? Thx\n>\n> Sent from my BlackBerry 10 smartphone.\n\n> **From:** Jared Birchall <jbirchall@muskfoundation.org>\n> **To:** Teresa Holland <[redacted]@ctcmycfo.com>\n> **Cc:** Ronald Gong <[redacted]@ctcmycfo.com>, Paula Lo <[redacted]@ctcmycfo.com>\n> **Subject:** Re: projected giving\n> **Date:** Friday, October 28, 2016 9:50 PM\n>\n> **I reconfirmed with Elon that he is ok with us moving shares from the Musk Foundation as needed to cover any charitable commitments. It looks like we should consider sending 26k shares, which would give us about $5.2MM to tide us over through the end of the year. Agreed?**\n>\n> Thanks,\n> Jared\n>\n> *Jared J. Birchall*\n> Managing Director\n> **Musk Family Office**\n\n> **From:** Teresa Holland <[redacted]@ctcmycfo.com>\n> **To:** Jared Birchall <jbirchall@muskfoundation.org>\n> **Subject:** projected giving\n> **Date:** Thu, Oct 27, 2016 at 3:59 PM\n>\n> Hi Jared —\n>\n> Please see the schedule for discussion of the liquidity needs for charitable giving, and the anticipated stock grant from the foundation to a DAF account during the upcoming window.\n>\n> The current DAF account balances are:\n>\n> The Musk Foundation Charitable Fund **$914,298**\n> The Musk Charitable Fund **$1,647,557**\n>\n> Many thanks —\n> Teresa\n>\n> Teresa A. Holland — Director, CTC | myCFO, LLC\n\n[Remainder is BMO Harris / Morgan Stanley confidentiality and Dodd-Frank Section 975 boilerplate — Bates EXMF-0003988–0003989.]\n\n## Commentary\n\nPX 84 is one of plaintiffs' best documentary supports for the **donor-advised-fund mechanics** of Musk's giving and Birchall's role as the man-on-the-spot financial intermediary, but it is also the very evidence the defense uses for its tax-deduction-wrinkle attack (see [[Key Themes]]: \"for every single one of those donations… Mr. Musk was entitled to take a tax deduction\"). The chain shows the operational pattern — Musk Foundation → DAF → grant to OpenAI — that Birchall confirmed on direct on [[Day 4|Day 4]] when Wilson cross-examined him on whether Musk could have unwound a DAF contribution (he could not). The \"26k shares ~ $5.2MM\" sizing here is consistent with Birchall's PX 112A reconstruction of the **$38,191,066** total in approximately 60 contributions across 2016–2020. Note that Birchall's \"I reconfirmed with Elon that he is ok with us moving shares\" is exactly the kind of contemporaneous principal-confirmation plaintiffs cite as evidence Musk personally signed off on the OpenAI grants — and exactly the kind of routine instruction defense cites as evidence these were unconditional charitable transfers, not contractually restricted donations.\n\n---\n*See also:* [[Day 4|Day 4 digest]] · [[Jared Birchall]] · [[PX-112A]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-86", "exhibit": "PX 86", "party": "Plaintiffs", "type": "Letter", "admitted_trial_day": "Day 4 (April 30, 2026)", "uploaded_box_pt": "2026-04-30T15:27:01", "uploader": "Someone", "pages": 1, "size_bytes": 198564, "source_pdf": "PX-86.pdf", "bates": "2024Foundation-0000001", "pdf_url": "https://media.mts-in.com/PX-86.pdf", "body_markdown": "# PX 86 — YC.org thank-you letter for $5M Musk Foundation gift (March 1, 2017)\n\n> Chris Clark's March 1, 2017 letter to Elon Musk on YC.org letterhead, gratefully acknowledging \"your generous gift of $5,000,000.00 from the Musk Foundation on February 27, 2017\" and confirming the gift \"qualifies as a charitable contribution for federal tax purposes.\"\n\n## Document type\n**Letter.** One-page tax-acknowledgement letter on YC.org letterhead (335 Pioneer Way, Mountain View, CA 94041), dated March 1, 2017, signed by Chris Clark, Treasurer, YC.org. Marked \"Confidential.\" Bates 2024Foundation-0000001 (also bearing handwritten \"H-04\").\n\n## Logistics\n- **Trial admission:** Day 4 (April 30, 2026) — used in the Birchall direct/cross sequence on Musk's contribution accounting and the donor-advised-fund/tax-deduction line of attack. See [[Key Themes]] §\"The 'donor-advised funds' / tax-deduction wrinkle\" — every donation in the chain entitled Musk to a tax deduction; this letter is the receipt for one of the larger ones.\n- **Box upload:** 2026-04-30 15:27:01 PT — uploaded thirteen seconds after PX 17 in the Day 4 cross-prep batch.\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~199 KB, 1 page.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `086.pdf`.\n\n## Transcribed text\n\n> YC.org\n> 335 Pioneer Way\n> Mountain View, CA 94041\n>\n> March 1, 2017\n>\n> Dear Mr. Musk:\n>\n> This letter is to gratefully acknowledge, for your tax records, our receipt of your generous gift of $5,000,000.00 from the Musk Foundation on February 27, 2017.\n>\n> No goods or services were provided to you in exchange for this gift. Therefore, the full amount of your gift qualifies as a charitable contribution for federal tax purposes. For reference, YC.org's EIN is: 47-3772053.\n>\n> Thank you again for your generous gift!\n>\n> Sincerely,\n>\n> *(signature: Christopher R Clark)*\n>\n> Chris Clark\n> Treasurer\n> YC.org\n\n## Commentary\n\nPX 86 is a small-but-load-bearing piece of plaintiffs' contribution-tracking case: the documentary acknowledgement of one tranche of the $38.19M total that Birchall walked the jury through on Day 4 (PX 112A — see [[Key Themes]]). The letter's significance is structural rather than narrative — it shows the gift went *to YC.org*, not directly to OpenAI Inc., which is the kind of routing detail plaintiffs use to dispute the defense's \"$1B / $38M gap\" narrative (\"the money flowed through DAFs and YC.org to OpenAI\") and which the defense uses for the converse point (Wilson's tax-deduction cross of Birchall: \"for every single one of those donations, it's somewhere in the chain, Mr. Musk was entitled to take a tax deduction\" — yes). YC.org (Y Combinator's nonprofit research arm, EIN 47-3772053) was Sam Altman's vehicle and the entity through which a meaningful slice of OpenAI's early funding moved before reaching OpenAI's own balance sheet — relevant to the Pioneer Building / lease story (see [[Key Themes]] §\"The Pioneer Building\"). Chris Clark, the signer, reappears as the author of the July 2020 OpenAI-pays-the-rent email plaintiffs use as the alleged 2020 charitable solicitation defeating the statute of limitations.\n\n---\n*See also:* [[Day 4|Day 4 digest]] · [[PX 17]] · [[PX 100]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-87", "exhibit": "PX 87", "party": "Plaintiffs", "type": "Letter (tax acknowledgment)", "admitted_trial_day": "Day 4 (April 30, 2026)", "uploaded_box_pt": "2026-04-30T15:27:05", "uploader": "Someone", "pages": 1, "size_bytes": 197743, "source_pdf": "PX-87.pdf", "pdf_url": "https://media.mts-in.com/PX-87.pdf", "body_markdown": "# PX 87 — June 1, 2017 YC.org tax-acknowledgment letter for $5M Musk Foundation gift\n\n> Chris Clark's one-page tax-receipt letter formally acknowledging a $5,000,000 May 26, 2017 contribution from the Musk Foundation as a charitable contribution to YC.org for federal tax purposes.\n\n## Document type\n**Letter (tax-substantiation receipt).** A single-page letter on YC.org letterhead (335 Pioneer Way, Mountain View, CA 94041), dated June 1, 2017, signed by Chris Clark in his capacity as Treasurer of YC.org. Bates 2024Foundation-0000002, marker H-05.\n\n## Logistics\n- **Trial admission:** Day 4 (April 30, 2026) — used during Birchall direct in support of the donation-by-donation walk-through that established the precise $38,191,066 figure (PX 112A) and the corresponding chain of tax deductions. No prior wiki reference; admission day inferred from Box upload date.\n- **Box upload:** 2026-04-30 15:27:05 PT — late-afternoon Day 4 batch (clustered with PX 71, PX 24, PX 103, PX 1504, DX 539, DX 646, DX 1156, DX 1285).\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~193 KB, 1 page.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `087.pdf`.\n\n## Transcribed text\n\n> YC.org\n> 335 Pioneer Way\n> Mountain View, CA 94041\n>\n> June 1, 2017\n>\n> Dear Mr. Musk:\n>\n> This letter is to gratefully acknowledge, for your tax records, our receipt of your generous gift of $5,000,000.00 from the Musk Foundation on May 26, 2017.\n>\n> No goods or services were provided to you in exchange for this gift. Therefore, the full amount of your gift qualifies as a charitable contribution for federal tax purposes. For reference, YC.org's EIN is: 47-3772053.\n>\n> Thank you again for your generous gift!\n>\n> Sincerely,\n> [signature]\n> Chris Clark\n> Treasurer\n> YC.org\n\n## Commentary\n\nPX 87 is the documentary backstop for two things plaintiffs needed on the record: that the May 26, 2017 wire was processed as a charitable contribution to a 501(c)(3) (YC.org, EIN 47-3772053 — see also [[PX 71]] for YC.org's IRS determination letter), and that \"no goods or services were provided\" — so the entire $5M qualified as a deduction. The defense uses these acknowledgments to drive the parallel \"tax-deduction every step of the way\" theme: Birchall conceded on cross-examination that Musk was entitled to take a tax deduction on every single contribution (see [[Key Themes]] § \"donor-advised funds / tax-deduction wrinkle\"). The 335 Pioneer Way address is also notable — it's the same Mountain View building YC and YC.org operated out of in 2015–2017, before OpenAI moved to the San Francisco Pioneer Building (the Musk Industries lease — see [[DX 539]]).\n\n---\n*See also:* [[Day 4|Day 4 digest]] · [[PX 71]] · [[Jared Birchall]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-90", "exhibit": "PX 90", "party": "Plaintiffs", "type": "Email", "admitted_trial_day": "Day 6 (May 4, 2026)", "uploaded_box_pt": "2026-05-04T14:46:05", "uploader": "Someone", "pages": 1, "size_bytes": 199536, "source_pdf": "PX-90.pdf", "pdf_url": "https://media.mts-in.com/PX-90.pdf", "body_markdown": "# PX 90 — June 28, 2017 Brockman → Musk \"Re: The 10,000\" (Microsoft / Azure GPU ask)\n\n> Musk's June 28, 2017 message reporting that he \"Just spoke to Satya\" about OpenAI needing \"the order of 10,000 servers with the latest Nvidia GPUs to beat the best human players at competitive eSports games\" — and Brockman's brief acknowledgment.\n\n## Document type\n**Email thread, plain text, two messages.** Greg Brockman's June 28, 2017 9:14 PM reply on top of Musk's same-day 2:10 PM original to Brockman, cc Sutskever and Altman. Subject \"Re: The 10,000.\" Bates OPENAI_MUSK00000308.\n\n## Logistics\n- **Trial admission:** Day 6 (May 4, 2026) — used during Brockman cross by Mr. Kry on the Microsoft thread / compute-supply line of questioning. No prior wiki reference; admission day inferred from Box upload date.\n- **Box upload:** 2026-05-04 14:46:05 PT — Day 6 early-afternoon batch (clustered with PX 359 — Charter 2.0).\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~195 KB, 1 page.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `090.pdf`.\n\n## Transcribed text\n\n> **From:** Greg Brockman <gdb@openai.com>\n> **To:** Elon Musk <erm@spacex.com>\n> **Cc:** Ilya Sutskever <ilyasu@openai.com>, Sam Altman <shga@ycombinator.com>\n> **Subject:** Re: The 10,000\n> **Date:** 6/28/2017 9:14:11 PM\n>\n> Awesome, thanks. We'll get back to you shortly.\n>\n> - gdb\n\n> **From:** Elon Musk <erm@spacex.com>\n> **To:** Greg Brockman <gdb@openai.com>\n> **Cc:** Ilya Sutskever <ilyasu@openai.com>, Sam Altman <shga@ycombinator.com>\n> **Subject:** The 10,000\n> **Date:** Wed, Jun 28, 2017 at 2:10 PM\n>\n> Just spoke to Satya and talked about OpenAI needing one the order of 10,000 servers with the latest Nvidia GPUs to beat the best human players at competitive esports games.\n>\n> This would obviously be a major opportunity for Microsoft to promote Azure relative to other cloud systems.\n>\n> He said he'd talk internally and get back to me soon. Please send me the ideal spec for the individual servers and how they are connected. What ideally do we want Microsoft to do? Sounds like there is a good chance they will do it.\n\n## Commentary\n\nPX 90 is an early data point in the [[Microsoft thread]]: Musk personally lobbying Satya Nadella for ~10,000 Azure-hosted Nvidia GPUs **eighteen months before** the eventual Microsoft–OpenAI partnership announcement. Plaintiffs use it to argue that Musk's contribution was not just dollars but the strategic Microsoft introduction that ultimately powered the for-profit conversion (a contribution defense cannot easily dollarize). The \"competitive esports games\" framing is a reference to OpenAI's Dota 1v1 demonstration that summer — the company's signature 2017 capability win. Brockman's three-word reply (\"Awesome, thanks\") is itself plaintiffs' point: he didn't have a Satya relationship of his own. Cross-reference: see [[DX 748]] for Brockman's January 2018 \"fundraising structure\" follow-on and [[PX 233|PX 233 / Microsoft]] for the long-arc Azure narrative.\n\n---\n*See also:* [[Day 6|Day 6 digest]] · [[Greg Brockman]] · [[Microsoft thread|Key Themes — Microsoft thread]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-91", "exhibit": "PX 91", "party": "Plaintiffs", "type": "Compendium / multi-doc", "admitted_trial_day": "Day 4 (April 30, 2026)", "uploaded_box_pt": "2026-04-30T15:27:32", "uploader": "Someone", "pages": 82, "size_bytes": 9423699, "source_pdf": "PX-91.pdf", "pdf_url": "https://media.mts-in.com/PX-91.pdf", "body_markdown": "# PX 91 — Fidelity Charitable grant confirmation letters from the Musk Foundation Charitable Fund to OpenAI Inc. (July 2017 – September 2020)\n\n> A compiled run of ~40 monthly Fidelity Charitable donor-advised-fund grant confirmation letters routing Musk's contributions to OpenAI Inc. through the Musk Foundation Charitable Fund (Giving Account ending 6490) — the documentary backbone of plaintiffs' \"$38 million contributed\" narrative and defense's \"tax-deduction every step of the way\" wrinkle.\n\n## Document type\n**Compendium / multi-doc.** Sequential Fidelity Charitable grant-confirmation letters, two pages each, on Fidelity Charitable letterhead from P.O. Box 770001, Cincinnati, OH 45277-0053. Each letter is addressed to Elon Musk personally and confirms a grant from the **Musk Foundation Charitable Fund, Giving Account ending 6490**, to **OPENAI INC** (originally 335 Pioneer Way, Mountain View, CA; later 3180 18th St Ste 100, San Francisco). Bates: FIDCHAR-OPENAI-000761 through FIDCHAR-OPENAI-000842.\n\n## Logistics\n- **Trial admission:** Day 4 (April 30, 2026) — Musk cross + redirect; Birchall direct + cross. The exhibit underlies Birchall's reading of the **\"approximately $38,191,066\"** figure in ~60 contributions for plaintiffs (PX 112A) — see [[Key Themes]] § \"The $1B / $38M gap.\"\n- **Box upload:** 2026-04-30 15:27:32 PT — Day 4 mid-afternoon batch.\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~9.4 MB, 82 pages (~40 grant-confirmation letters at two pages apiece).\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `091.pdf`.\n\n## Transcribed text\n\nEach letter follows the same template. A representative example (the first in the sequence):\n\n> **FIDELITY Charitable**\n> P.O. Box 770001, Cincinnati, OH 45277-0053\n>\n> ELON MUSK\n> [redacted street address]\n> LOS ANGELES, CA 90049\n>\n> July 18, 2017\n>\n> Dear Elon Musk:\n>\n> We are pleased to inform you that, at your recommendation, Fidelity Charitable has made a grant from the Musk Foundation Charitable Fund, Giving Account [redacted]6490, to the following grant recipient(s):\n>\n> **Grant ID:** 9178019\n> **Check Number:** 6549722\n> **Amount of Grant:** $175,000.00\n> **Source:** Giving Account\n> **Grant Recipient's Name:** OPENAI INC\n> **Grant Recipient's Address:** 335 PIONEER WAY, MOUNTAIN VIEW, CA 94041-1505\n> **Special Purpose:** Designation: General Support\n> **Acknowledge To:** Anonymous Donor\n>\n> *Giving Account*\n> 07/18/17 — Money Market — (11,097.020) units — $15.77 — $(175,000.00) — Unit Balance 236,062.302\n> Total Grant Amount: $(175,000.00)\n> Total Giving Account Balance: $3,722,702.51\n\n**Tabulated grant ledger (extracted from each successive letter):**\n\n| Letter date | Amount | Recipient address | Acknowledge To |\n|---|---:|---|---|\n| Jul 18, 2017 | $175,000 | 335 Pioneer Way, Mountain View | Anonymous Donor |\n| Aug 14, 2017 | $175,000 | Mountain View | Anonymous Donor |\n| Sep 15, 2017 | $175,000 | Mountain View | Anonymous Donor |\n| Sep 29, 2017 | $85,000 | Mountain View | Anonymous (only account name disclosed) |\n| Oct 16, 2017 | $235,000 | Mountain View | Anonymous Donor |\n| Nov 14, 2017 | $235,000 | Mountain View | Anonymous Donor |\n| Dec 14, 2017 | $235,000 | Mountain View | Anonymous Donor |\n| Jan 18, 2018 | $290,000 | Mountain View | Anonymous Donor |\n| Feb 20, 2018 | $390,000 | Mountain View | Anonymous Donor |\n| Mar 14, 2018 | $290,000 | Mountain View | Anonymous Donor |\n| Apr 16, 2018 | $290,000 | Mountain View | Anonymous Donor |\n| May 15, 2018 | $290,000 | Mountain View | Anonymous Donor |\n| Jun 14, 2018 | $290,000 | Mountain View | Anonymous Donor |\n| Jul 16, 2018 | $290,000 | Mountain View | Anonymous Donor |\n| Aug 14, 2018 | $290,000 | Mountain View | Anonymous Donor |\n| Sep 18, 2018 | $290,000 | Mountain View | Anonymous Donor |\n| Oct 17, 2018 | $290,000 | Mountain View | Anonymous Donor |\n| Nov 14, 2018 | $290,000 | Mountain View | Anonymous Donor |\n| Dec 17, 2018 | $290,000 | Mountain View | Anonymous Donor |\n| Jan 16, 2019 | $290,000 | Mountain View | Anonymous Donor |\n| Feb 14, 2019 | $290,000 | Mountain View | Anonymous Donor |\n| Mar 22, 2019 | $290,000 | Mountain View | Anonymous Donor |\n| Apr 16, 2019 | $290,000 | Mountain View | Anonymous Donor |\n| May 14, 2019 | $290,000 | Mountain View | Anonymous Donor |\n| Jun 14, 2019 | $290,000 | Mountain View | Anonymous Donor |\n| Jul 17, 2019 | $290,000 | Mountain View | Anonymous Donor |\n| Aug 14, 2019 | $290,000 | Mountain View | Anonymous Donor |\n| Sep 16, 2019 | $290,000 | Mountain View | Anonymous Donor |\n| Oct 17, 2019 | $290,000 | Mountain View | Anonymous Donor |\n| Nov 15, 2019 | $290,000 | 3180 18th St, San Francisco | Anonymous Donor |\n| Dec 17, 2019 | $290,000 | San Francisco | Anonymous Donor |\n| Jan 14, 2020 | $290,000 | San Francisco | Anonymous Donor |\n| Feb 14, 2020 | $290,000 | San Francisco | Anonymous Donor |\n| Mar 16, 2020 | $290,000 | San Francisco | Anonymous Donor |\n| Apr 13, 2020 | $290,000 | San Francisco | Anonymous Donor |\n| May 13, 2020 | $290,000 | San Francisco | Anonymous Donor |\n| Jun 15, 2020 | $290,000 | San Francisco | Anonymous Donor |\n| Jul 14, 2020 | $290,000 | San Francisco | Anonymous Donor |\n| Aug 17, 2020 | $290,000 | San Francisco | Anonymous Donor |\n| Sep 14, 2020 | $290,000 | San Francisco | Anonymous Donor |\n\nEach letter carries the standard Fidelity boilerplate at the foot:\n\n> Please note that this correspondence confirming the grant(s) is for your reference only. **It does not reflect an additional tax-deductible contribution made by you and cannot be used for tax purposes.** … Fidelity Charitable is the brand name for Fidelity Charitable Gift Fund, an independent public charity with a donor-advised fund program.\n\nThe address-line change in late-2019 letters (Mountain View → 3180 18th St Ste 100, San Francisco) tracks OpenAI's move to the Pioneer Building in San Francisco — see [[Key Themes]] § \"The Pioneer Building.\" Several letters (Sep 18, 2018; Dec 17, 2018; Mar 22, 2019; Oct 17, 2019; Dec 17, 2019; Jul 14, 2020) include redacted black-box passages that obscure the recommended grant write-up and a parallel grant amount on the same date (e.g., Sep 18, 2018 shows total $315,000 sold; Dec 17, 2018 total $340,000; Mar 22, 2019 total $340,000; Oct 17, 2019 total $310,000; Dec 17, 2019 total $385,000; Jul 14, 2020 total $400,000) — i.e., contemporaneous co-grants to other recipients on the same day, redacted from the produced exhibit.\n\n## Commentary\n\nPX 91 is the documentary spine of plaintiffs' \"Musk paid for OpenAI\" story and, at the same time, of the defense's \"donor-advised funds\" counter-attack. **Plaintiffs' read:** here are the actual checks, dollar by dollar, month by month — the receipts behind Birchall's testimony that Musk supplied \"approximately 60 percent\" of OpenAI's 2016–2017 contributions while Brockman contributed \"zero\" (see [[Key Themes]]). **Defense's read:** every one of these forty-odd grants moved through Fidelity Charitable's donor-advised fund — meaning Musk took a charitable tax deduction at the moment of the original Fidelity contribution, not the moment of the OpenAI grant; once the funds entered Fidelity, \"he could not take the money back\" (see [[Key Themes]] § \"donor-advised funds / tax-deduction wrinkle\"; [[Day 4|Day 4 digest]] for Birchall's cross by Wilson on the DAF point). The recipient-address change from Mountain View to **3180 18th St Ste 100, San Francisco** in the November 2019 letter dates OpenAI's move into the Pioneer Building — relevant to the [[Key Themes]] § \"The Pioneer Building\" line of evidence. The \"Acknowledge To: **Anonymous Donor**\" line in every letter is a quiet but important detail: Musk's contributions to OpenAI ran through Fidelity under anonymous designation, complicating any later \"I publicly funded the charity\" narrative.\n\n---\n*See also:* [[Day 4|Day 4 digest]] · [[PX 112A]] · [[Jared Birchall]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-92", "exhibit": "PX 92", "party": "Plaintiffs", "type": "Email thread", "admitted_trial_day": "Day 4 (April 30, 2026)", "uploaded_box_pt": "2026-04-30T15:27:07", "uploader": "Someone", "pages": 1, "size_bytes": 101917, "source_pdf": "PX-92.pdf", "pdf_url": "https://media.mts-in.com/PX-92.pdf", "body_markdown": "# PX 92 — July 13–19, 2017 YCR Basic Income Grant ($250K Musk → YC.org UBI Project)\n\n> A one-page email chain in which Chris Clark (YC.org) solicits a Musk Foundation contribution to YC Research's Universal Basic Income Project, Birchall instructs UBS to process a $250,000 grant from Musk's Fidelity account, and UBS confirms the check will go out the following day — a separate Musk-to-YC.org gift outside the $5M and $16M tracked OpenAI lines.\n\n## Document type\n**Email thread, plain text, single page.** Three messages: Chris Clark (chris@ycr.org) → Birchall on July 13, 2017; Birchall (jbirchall@muskfoundation.org) → Leeder Hsu and Aimee Harries at UBS on July 18, 2017 8:36 PM; Leeder Hsu (leeder.hsu@ubs.com) → Birchall confirmation, July 19, 2017 7:01:57 PM. Bates EXMF-0000102. Stamped \"Confidential.\" Several UBS staff email addresses redacted in black overlays.\n\n## Logistics\n- **Trial admission:** Day 4 (April 30, 2026). Used during Birchall direct on the donation-by-donation walk-through. The exhibit also touches on plaintiffs' [[Key Themes]] § \"donor-advised funds / tax-deduction wrinkle\" because the gift is processed via Fidelity (a DAF custodian) and the recipient (YC.org) is the same 501(c)(3) ([[PX-71]]) referenced in the [[PX-87]] tax acknowledgment.\n- **Box upload:** 2026-04-30 15:27:07 PT — late-afternoon Day 4 batch (clustered with PX 52, PX 75, PX 87).\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~99 KB, 1 page.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `092.pdf`.\n\n## Transcribed text\n\n> **From:** Leeder Hsu <leeder.hsu@ubs.com>\n> **Sent:** 7/19/2017 7:01:57 PM\n> **To:** Jared Birchall <jbirchall@muskfoundation.org>\n> **Cc:** [redacted]@ubs.com; [redacted]@ubs.com\n> **Subject:** RE: YCR Basic Income Grant\n> **Attachments:** disclaim.txt\n>\n> Hi Jared,\n>\n> I just confirmed with Fidelity that the grant has been approved and is set to go out by check tomorrow.\n>\n> Let us know if we can help with anything else.\n>\n> Thanks,\n> Leeder\n\n> **From:** Jared Birchall <jbirchall@muskfoundation.org>\n> **To:** Leeder Hsu <leeder.hsu@ubs.com>, Aimee Harries\n> **Subject:** Fwd: YCR Basic Income Grant\n> **Date:** Tuesday, July 18, 2017 8:36 PM\n>\n> Hi Leeder,\n>\n> Can you please process the grant below?\n>\n> $250k to YC.org. As noted, please include a note that this is for the benefit of the UBI study. No mega urgency here. A check is fine.\n>\n> Thanks!\n> Jared\n\n> **From:** Chris Clark <chris@ycr.org>\n> **To:** Jared Birchall <jbirchall@muskfoundation.org>\n> **Cc:** Lindsay Roitman <lindsay@ycr.org>\n> **Subject:** YCR UBI\n> **Date:** Thu, Jul 13, 2017 at 7:43 PM\n>\n> Hi Jared,\n>\n> Lindsay mentioned that Elon is interested in donating to YCR's UBI Project. I've attached the determination letter and wire instructions for YC ORG, which is the 501(c)(3) that issues the UBI grants.\n>\n> If you have the ability to include a memo in the transfer it would be helpful to put something like \"For the benefit of the UBI Project.\" We'll make sure it gets credited to that project regardless, but it's helpful to have that associated with the transfer for our records.\n>\n> Let me know if you have any questions or need anything else.\n>\n> Chris\n\n## Commentary\n\nPX 92 is a small but useful exhibit on three Day 4 themes. First, it confirms that **Chris Clark** — the same OpenAI Inc. director/COO/treasurer who signed the [[PX-52]] 2017 Form 990 and the [[PX-87]] tax acknowledgment letter — was simultaneously soliciting Musk-side gifts on behalf of Y Combinator Research (YC.org) for non-OpenAI projects (the UBI Project), the same entity that later flowed the $16M Musk Foundation grant to OpenAI Inc. (see [[PX-52]] Schedule B contributor #5). Second, Birchall's use of **Fidelity** as the disbursing custodian, with UBS as the relationship-management front end, fits the pattern Birchall conceded on cross — every Musk gift was funneled through a vehicle that produced a tax deduction at the source. Third, the $250K UBI gift itself is **not** captured in the $38,191,066 OpenAI total that Birchall recited from PX 112A — plaintiffs use this to set up the larger architecture in which Musk's \"charitable giving in this ecosystem\" was substantially larger than OpenAI alone, while defense uses it to argue that Musk's relationship with YC.org/YC Research extended beyond OpenAI and was driven by Sam Altman's broader philanthropic agenda (UBI, OpenAI, YC Research). The \"no mega urgency here. A check is fine\" tone is also useful color for the family-office-mechanics theme in [[Jared Birchall]] cross.\n\n---\n*See also:* [[Day 4|Day 4 digest]] · [[PX-52]] · [[PX-71]] · [[PX-87]] · [[Jared Birchall]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-921", "exhibit": "PX 921", "party": "Plaintiffs", "type": "Blog post", "admitted_trial_day": "Day 4 (April 30, 2026)", "uploaded_box_pt": "2026-04-30T14:42:54", "uploader": "Someone", "pages": "unknown", "size_bytes": 1116547, "source_pdf": "PX-921.pdf", "pdf_url": "https://media.mts-in.com/PX-921.pdf", "body_markdown": "# PX 921 — Sept 22, 2020 OpenAI blog post: \"OpenAI licenses GPT-3 technology to Microsoft\"\n\n> The public blog post on **September 22, 2020** in which OpenAI announced — two days before Musk's \"**OpenAI is essentially captured by Microsoft**\" tweet — that \"**OpenAI has agreed to license GPT-3 to Microsoft for their own products and services.**\"\n\n> *Note on stamping:* the file `921.pdf` carries a **DX-0921** Defendant exhibit stamp, but the chunk-5 TSV catalogues this as **PX 921** (Plaintiffs' exhibit) per the Box upload metadata. The discrepancy may reflect dual-listing or a clerical inconsistency at admission; the canonical filename in this wiki follows the TSV.\n\n## Document type\n**Blog post** — OpenAI.com news/product post, dated September 22, 2020, captioned \"Product.\" Bates OPENAI_MUSK00039065 onward. Multi-page screenshot capture; first page contains the headline, lead, and three opening paragraphs.\n\n## Logistics\n- **Trial admission:** Day 4 (April 30, 2026) — Musk cross + redirect; Birchall direct + cross. Used as the **public predicate** for Musk's September 24, 2020 \"captured by Microsoft\" tweet (PX 251 / DX evidence — see [[Key Themes]] §\"Captured by Microsoft (statute-of-limitations fight)\").\n- **Box upload:** 2026-04-30 14:42:54 PT — Day 4 early-afternoon Plaintiffs' batch (uploaded ~45 minutes before [[PX 60]], [[PX 79]], [[PX 93]]).\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~1.1 MB, multiple pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `921.pdf`.\n\n## Transcribed text\n\n> **September 22, 2020 — Product**\n>\n> # OpenAI licenses GPT-3 technology to Microsoft\n>\n> OpenAI has agreed to license GPT-3 to Microsoft for their own products and services.\n>\n> [Image: an Impressionist-style landscape painting]\n>\n> OpenAI released its first commercial product back in June: an API for developers to access advanced technologies for building new applications and services. The API features a powerful general purpose language model, **GPT-3** (opens in a new window), and has received tens of thousands of applications to date.\n>\n> In addition to offering GPT-3 and future models via the OpenAI API, and as part of a multiyear partnership (opens in a new window) announced last year, **OpenAI has agreed to license GPT-3 to Microsoft for their own products and services**. The deal has no impact on continued access to the GPT-3 model through OpenAI's API, and existing and future users of it will continue building applications with our API as usual.\n>\n> Unlike most AI systems which are designed for one use-case, OpenAI's API today provides a general-purpose \"text in, text out\" interface, allowing users to try it on virtually any English language task. **GPT-3 is the most powerful model behind the API today, with 175 billion parameters.** There are several other models available via the API today, as well as other technologies and filters that allow developers to customize GPT-3 and other language models for their own use.\n>\n> [Subsequent pages of the PDF continue the September 22, 2020 OpenAI blog post — additional sections describing the API, GPT-3, the Microsoft licensing arrangement, and the multiyear partnership context. Verbatim transcription of subsequent pages omitted from this entry.]\n\n## Commentary\n\nPX 921 is the **public-facing artifact** that triggers Musk's September 24, 2020 \"**This does seem like the opposite of open. OpenAI is essentially captured by Microsoft.**\" tweet — defense's most-cited statute-of-limitations exhibit (see [[Key Themes]] §\"Captured by Microsoft\"). Plaintiffs' use of the same blog post on direct is to anchor the moment in time at which the **commercial-IP licensing of an OpenAI flagship model to Microsoft** stopped being a private commercial-negotiation document (cf. [[PX 244]], the May 2019 Murati ↔ Waymouth Odyssey-JDCA email) and became a public, irrevocable fact — and to map the \"we plan to open source the current version\" undertaking in PX 244 against the actual September 2020 path. The blog post is also the textual link from the 2019 Microsoft round to the 2023 JDCA's \"all IP and technology … excluding only AGI\" carveout that Wu authenticated on Day 7 (see [[Key Themes]] §\"The 2023 JDCA\"). Dual-stamping note: the PDF carries a DX-0921 sticker but the chunk-5 metadata catalogues the file as PX 921; the discrepancy may indicate a dual-listed exhibit or a clerical inconsistency.\n\n---\n*See also:* [[Day 4|Day 4 digest]] · [[PX 244]] · [[Key Themes]] · [[Sam Altman]] · [[Elon Musk]]\n"}
{"exhibit_id": "PX-93", "exhibit": "PX 93", "party": "Plaintiffs", "type": "Email thread", "admitted_trial_day": "Day 4 (April 30, 2026)", "uploaded_box_pt": "2026-04-30T15:26:46", "uploader": "Someone", "pages": 2, "size_bytes": 265794, "source_pdf": "PX-93.pdf", "pdf_url": "https://media.mts-in.com/PX-93.pdf", "body_markdown": "# PX 93 — July 21–22, 2017 Musk / Brockman / Sutskever / Schulman / Zaremba thread on China + the path to \"AI research + hardware for-profit (starting 2018)\"\n\n> Brockman's contemporaneous \"100% agreed. We think the path must be: 1. AI research non-profit (through end of 2017) 2. AI research + hardware for-profit (starting 2018) 3. Government project (when: ??)\" — and Musk's same-day offer of Founder Series Tesla Model 3s as thank-you gifts for the OpenAI team.\n\n## Document type\n**Email thread, plain text, six messages** stitched into a single Birchall reply. Recipients: Musk, Sutskever, Brockman, Schulman, Zaremba; cc to Birchall on the final reply (\"Will do. Guys — I'll reach out separately with details.\"). Subject: \"Re: Beijing Wants A.I. to Be Made in China by 2030 — NYTimes.com.\" Bates 2024MUSK-0006126–0006127.\n\n## Logistics\n- **Trial admission:** Day 4 (April 30, 2026) — Musk cross + redirect; Birchall direct + cross. Slotted in the same Day 4 afternoon batch as [[PX 60]] and [[PX 79]].\n- **Box upload:** 2026-04-30 15:26:46 PT — Day 4 mid-afternoon Plaintiffs' batch (uploaded a few seconds before PX 60 and PX 79).\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~260 KB, 2 pages.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `093.pdf`.\n\n## Transcribed text\n\n> **From:** Jared Birchall <jbirchall@muskfoundation.org>\n> **To:** Elon Musk <erm@spacex.com>\n> **Cc:** Ilya Sutskever <ilyasu@openai.com>, Greg Brockman <gdb@openai.com>, John Schulman <joschu@openai.com>, Wojciech Zaremba <woj@openai.com>\n> **Subject:** Re: Beijing Wants A.I. to Be Made in China by 2030 — NYTimes.com\n> **Date:** Fri, 21 Jul 2017 13:56:06 -0700\n>\n> Will do.\n>\n> Guys - I'll reach out separately with details.\n>\n> Regards,\n> Jared\n\n> **From:** Elon Musk <erm@spacex.com>\n> **To:** Jared Birchall <jbirchall@muskfoundation.org>\n> **Cc:** Ilya Sutskever <ilyasu@openai.com>, Greg Brockman <gdb@openai.com>, John Schulman <joschu@openai.com>, Wojciech Zaremba <woj@openai.com>\n> **Subject:** Re: Beijing Wants A.I. to Be Made in China by 2030 — NYTimes.com\n> **Date:** Fri, Jul 21, 2017 at 1:53 PM\n>\n> Consider it done!\n>\n> Jared, please facilitate. Let's try to slot these four cars into next month's production.\n\n> **From:** Ilya Sutskever <ilyasu@openai.com>\n> **To:** Elon Musk <erm@spacex.com>\n> **Cc:** Greg Brockman <gdb@openai.com>, John Schulman <joschu@openai.com>, Wojciech Zaremba <woj@openai.com>\n> **Subject:** Re: Beijing Wants A.I. to Be Made in China by 2030 — NYTimes.com\n> **Date:** Sat, Jul 22, 2017 at 6:48 AM\n>\n> Wow, thank you! This is a very kind gesture that we greatly appreciate. Yes, Wojciech definitely deserves it as much as we do, as does +John Schulman.\n\n> **From:** Elon Musk <erm@spacex.com>\n> **To:** Ilya Sutskever <ilyasu@openai.com>, Greg Brockman <gdb@openai.com>\n> **Subject:** Re: Beijing Wants A.I. to Be Made in China by 2030 — NYTimes.com\n> **Date:** Fri, Jul 21, 2017 at 1:25 PM\n>\n> By the way, in appreciation for what you've done to get OpenAI to where it is today, I would like to give you each a Founder Series Model 3. These are the earliest cars produced and are not available to the public.\n>\n> If there are a few others at OpenAI who you think also really deserve one (maybe Wojciech?), I am happy to do the same for them.\n>\n> Hope you like it!\n>\n> Elon\n\n> **From:** Greg Brockman <gdb@openai.com>\n> **To:** Elon Musk <erm@spacex.com>\n> **Cc:** Ilya Sutskever <ilyasu@openai.com>\n> **Subject:** Re: Beijing Wants A.I. to Be Made in China by 2030 — NYTimes.com\n> **Date:** Sat, Jul 22, 2017 at 4:32 AM\n>\n> 100% agreed. We think the path must be:\n>\n> 1. AI research non-profit (through end of 2017)\n> 2. AI research + hardware for-profit (starting 2018)\n> 3. Government project (when: ??)\n>\n> Government project is helpful for financial resources, but crucial for security — don't know how to defend against Putin or North Korea otherwise. When ready, we must proactively form the project around us, rather than being slurped in against our will.\n>\n> - gdb\n\n> **From:** Elon Musk <erm@spacex.com>\n> **To:** Greg Brockman <gdb@openai.com>, Ilya Sutskever <ilyasu@openai.com>\n> **Subject:** Beijing Wants A.I. to Be Made in China by 2030 — NYTimes.com\n> **Date:** Fri, Jul 21, 2017 at 3:34 AM\n>\n> They will do whatever it takes to obtain what we develop. Maybe another reason to change course.\n>\n> https://mobile.nytimes.com/2017/07/20/business/china-artificial-intelligence.html\n\n## Commentary\n\nPX 93 is plaintiffs' clean evidence that, by **late July 2017**, Brockman was already telling Musk in writing that the path \"must be\" a **for-profit starting 2018** — six weeks before Brockman's August 21, 2017 \"this is the only chance we have to get out from Elon\" journal entry in [[PX 151]] and two months before the [[PX 157]] \"final straw\" email. Defense's read is symmetric and equally pointed: Musk's \"**Consider it done!**\" reply on the same thread is him signing off on for-profit hardware as the next step, and his $1.25M Tesla-gift gesture (four Founder Series Model 3s) is contemporaneous personal generosity that complicates the \"stole-a-charity\" framing. The \"Putin or North Korea\" / \"**slurped in against our will**\" exchange foreshadows Sutskever's September 20, 2017 \"Honest Thoughts\" letter (the body of [[PX 157]]), which warns that \"**the goal of OpenAI is to … avoid an AGI dictatorship**.\" See [[Brockman Journal]] for the Aug–Sept 2017 internal-to-OpenAI side of the same conversation.\n\n---\n*See also:* [[Day 4|Day 4 digest]] · [[PX 151]] · [[PX 152]] · [[PX 157]] · [[Brockman Journal]] · [[Greg Brockman]] · [[Ilya Sutskever]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-98", "exhibit": "PX 98", "party": "Plaintiffs", "type": "Email", "admitted_trial_day": "Day 3 (April 29, 2026)", "uploaded_box_pt": "2026-04-29T15:09:20", "uploader": "Someone", "pages": 1, "size_bytes": 207911, "source_pdf": "PX-98.pdf", "pdf_url": "https://media.mts-in.com/PX-98.pdf", "body_markdown": "# PX 98 — Jan 1, 2018 Sutskever → Musk, \"most overwhelmingly competent person in the world\"\n\n> Ilya Sutskever's New Year's Day 2018 thank-you note to Elon Musk (BCC: Greg Brockman) — written near the two-year mark of OpenAI — calling Musk \"the most overwhelmingly competent person in the world.\"\n\n## Document type\n**Email, plain text, single message.** From Ilya Sutskever <ilyasu@openai.com> to Elon Musk <erm@spacex.com>, BCC Greg Brockman <gdb@openai.com>, sent 1/1/2018 8:45:15 AM, subject \"Note.\" Bates OPENAI_MUSK00007198.\n\n## Logistics\n- **Trial admission:** Day 3 (April 29, 2026 — Musk direct concludes; cross begins). Per the Day 3 transcript table: \"PX 98 | Jan. 1, 2018 email Sutskever → Musk ('most overwhelmingly competent person in the world')\" `(042926TT.txt:1075–1106)`.\n- **Box upload:** 2026-04-29 15:09:20 PT — Day 3 mid-afternoon batch (clustered with PX 157, PX 233, PX 251, PX 296, PX 355).\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~203 KB, 1 page.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `098.pdf`.\n\n## Transcribed text\n\n> **From:** Ilya Sutskever <ilyasu@openai.com>\n> **To:** Elon Musk <erm@spacex.com>\n> **Bcc:** Greg Brockman <gdb@openai.com>\n> **Subject:** Note\n> **Date:** 1/1/2018 8:45:15 AM\n>\n> Hi Elon,\n>\n> We've been working together for almost exactly two years, so it felt appropriate to send this note.\n>\n> I enjoy working together. You quickly pushed me out of my academic comfort zone. With time, I grew to appreciate the vast depth of your strategic insight — I can think of many examples, the biggest one is the decision to build the custom hardware. I've made my share of non-negligible mistakes in 2017, but their lessons will help us overcome the harder challenges of the future.\n>\n> We're trying to accomplish a difficult task, but with effort, we may be able to do it. **It helps that we have the most overwhelmingly competent person in the world helping us.**\n>\n> Happy 2018. Let's make 2017 pale in comparison.\n\n## Commentary\n\nPX 98 is a small, surgical plaintiffs' exhibit: the OpenAI co-founder who would later, in November 2023, lead the board vote to fire Sam Altman is on the record in **January 2018** calling Musk \"**the most overwhelmingly competent person in the world**\" and crediting his \"vast depth of strategic insight.\" The BCC to Brockman — but not to Altman — is itself testimony. Plaintiffs use the email to support the \"**Musk-was-the-mission**\" framing: at the two-year mark, internally to OpenAI's two technical leaders, Musk was the indispensable figure, not a peripheral donor. It is also context for the [[Brockman Journal|August 21, 2017 journal entry]] in which Brockman wrote, three months earlier, that getting \"out from Elon\" was \"the only chance we have\" — i.e., the same person Sutskever celebrates in PX 98 was being privately characterized by Brockman as the obstacle to be removed. See [[Day 3|Day 3 digest]] at 042926TT.txt:1075–1106 for the testimony walkthrough.\n\n---\n*See also:* [[Day 3|Day 3 digest]] · [[Ilya Sutskever]] · [[Elon Musk]] · [[Greg Brockman]] · [[Brockman Journal]] · [[Key Themes]]\n"}
{"exhibit_id": "PX-99", "exhibit": "PX 99", "party": "Plaintiffs", "type": "Email", "admitted_trial_day": "Day 3 (April 29, 2026)", "uploaded_box_pt": "2026-04-29T15:09:19", "uploader": "Someone", "pages": 1, "size_bytes": 186058, "source_pdf": "PX-99.pdf", "pdf_url": "https://media.mts-in.com/PX-99.pdf", "body_markdown": "# PX 99 — Jan 1, 2018 Brockman \"thank you\" email to Musk (\"Let's make sure that the post-AGI world is one that is good for humans\")\n\n> Greg Brockman's New Year's Day 2018 thank-you note to Musk three months after the [[PX-157|\"final straw\" email]] — the post-reconciliation document plaintiffs use to show Brockman publicly professed devotion to the mission while privately writing the [[PX-154|journal]] passages plaintiffs cite for \"fire Elon\" and \"morally bankrupt.\"\n\n## Document type\n**Email, single message, plain text.** From Greg Brockman (gdb@openai.com) to Elon Musk (erm@openai.com), bcc Ilya Sutskever (ilyasu@openai.com). Bates OPENAI_MUSK00000814.\n\n## Logistics\n- **Trial admission:** Day 3 (April 29, 2026) — Musk direct concludes; cross begins.\n- **Box upload:** 2026-04-29 15:09:19 PT — Day 3 mid-afternoon batch (clustered with PX 157, PX 233, PX 296, PX 355).\n- **Uploader:** \"Someone\" (anonymized — likely plaintiffs' counsel, MoloLamken).\n- **File size:** ~182 KB, 1 page.\n- **Source:** Public Box folder per Pretrial Order No. 1, Dkt. 446 — file `099.pdf`.\n\n## Transcribed text\n\n> **From:** Greg Brockman <gdb@openai.com>\n> **To:** Elon Musk <erm@openai.com>\n> **Bcc:** Ilya Sutskever <ilyasu@openai.com>\n> **Subject:** thank you\n> **Date:** 1/1/2018 7:56:57 AM\n>\n> Hi Elon,\n>\n> We've been working together for 2 years now, and in every meeting with you I continue to learn, grow, and see the world in a new way. I particularly admire your clarity of purpose, the simple and consistent heuristics you use to evaluate decisions, and that you stick to what's right rather than what's easy.\n>\n> I am grateful and humbled that you see potential in Ilya and me and are willing to give us a chance. 2017 was a hard year, and we've made mistakes, but I believe the lessons we've learned will ultimately make us successful in our mission.\n>\n> It's an honor to work alongside you. Let's make sure that the post-AGI world is one that is good for humans.\n>\n> - gdb\n\n## Commentary\n\nPX 99 is a **public-facing reset** following the [[PX-157|September 20, 2017 \"final straw\" rupture]] and the [[PX-154|Brockman journal]] passages of August–November 2017. Brockman's \"**Let's make sure that the post-AGI world is one that is good for humans**\" closing is plaintiffs' Exhibit A on Brockman's *outward* mission alignment — the sentiment plaintiffs juxtapose with his Nov 5–6, 2017 journal entries discussed on [[Day 6]] and [[Day 7]] (\"Real decision is fire Elon\"; \"It'd be wrong to steal the nonprofit from him… they'd be pretty morally bankrupt\") and his Nov 12, 2017 \"**It would be nice to be making the billions**\" entry. Mr. Molo's recross frame on Day 7 — \"**It's just an after-the-fact manufactured excuse for the very honest emotion that you expressed in the journal at the time**\" — is built on this contrast: what Brockman wrote *to Musk* versus what he wrote *for himself*. The \"2017 was a hard year, and we've made mistakes\" line is also the textual antecedent for plaintiffs' theory that Brockman knew, at the moment he wrote this, that he was concealing the for-profit-without-Musk plan documented in his own contemporaneous diary.\n\n---\n*See also:* [[Day 3|Day 3 digest]] · [[PX-157]] · [[PX-154]] · [[Greg Brockman]] · [[Brockman Journal]] · [[Key Themes]]\n"}