Elon Musk spent Wednesday in a California federal court accusing Sam Altman and OpenAI's other cofounders of having "stole a charity," and left the stand having admitted under oath that Tesla is not currently pursuing artificial general intelligence — a direct contradiction of an X post he published just weeks earlier claiming "Tesla will be one of the companies to make AGI." He also conceded that a separate post boasting of a $100 million investment in OpenAI overstated the actual figure, $38 million, by a factor of nearly three. Musk argued his reputation and network made up the difference.
The case, before Judge Yvonne Gonzalez Rogers, alleges that Altman, Ilya Sutskever, Greg Brockman and the other cofounders tricked Musk into bankrolling a nonprofit before letting a for-profit arm dominate the lab. After hours of testy testimony, the verdict may hinge on a narrow question: how much weight jurors give to the distinction between OpenAI investors whose profits are capped and those whose profits are not.
Microsoft's earliest major investments came with profit caps. Those restrictions have been progressively rolled back, and Musk testified that the unwinding of those caps is what finally pushed him to sue. In his telling, the cofounders went from building AI for humanity to "looting the nonprofit."
Key facts
- 01Musk admitted under oath that Tesla is not pursuing AGI, contradicting an X post weeks earlier claiming Tesla would be one of the companies to build it.
- 02He conceded his public claim of investing $100 million in OpenAI overstated the $38 million he actually contributed.
- 03Musk testified he discussed converting OpenAI to for-profit as early as 2016 and explored a 2017 plan giving himself majority equity.
- 04He stopped regular donations after those plans collapsed but kept paying for OpenAI's office space until 2020.
- 05Judge Yvonne Gonzalez Rogers ruled xAI and OpenAI's safety approaches are fair game for cross-examination Thursday.
OpenAI's lawyer William Savitt spent cross-examination complicating that arc. Savitt walked Musk through his own history of pushing OpenAI toward a for-profit structure, including discussions about folding the lab into Tesla. Musk acknowledged he raised converting OpenAI to a for-profit as early as 2016, and that in 2017 he explored creating a for-profit arm in which he would hold the majority of the equity and control the company.
“Musk testified he had claimed a $100 million investment in OpenAI on X when only $38 million actually changed hands, arguing his reputation and network made up for the gap.”— Jaeden Schafer
When those plans collapsed, Musk stopped making regular donations but continued paying for OpenAI's office space until 2020. Savitt also established that Musk had been consulted by Altman and Shivon Zillis — Musk's longtime adviser, mother of four of his children, and a former OpenAI board member — about later fundraising rounds, and did not object at the time. Zillis was on the board when some of those transactions were approved.
The cross-examination then turned to Tesla. After Musk testified that Tesla's AI work was focused only on self-driving rather than AGI, Savitt produced the recent X post claiming the opposite. "We are not pursuing AGI right now," Musk told the court — a statement Tesla shareholders, who have been sold an AGI narrative for years, may want to read carefully.
Savitt also surfaced emails showing Musk had backed efforts by Tesla and Neuralink to poach OpenAI staff while he was still on OpenAI's board. One thread concerned Andrej Karpathy, who left OpenAI to lead self-driving at Tesla. Another involved Sutskever, whom Zillis suggested Musk try to recruit to Tesla.
The most consequential exchange may have been about safety, the foundation of Musk's argument that OpenAI's corporate transition is dangerous to society. Savitt got Musk to concede that every AI company, xAI included, faces the same commercial pressures that erode safety focus. Gonzalez Rogers halted that line for the day but signaled it would resume Thursday with limits.
When Musk's lawyers tried to introduce ChatGPT's alleged role in the 2024 Tumbler Ridge shooting in Canada — an incident now the subject of seven California lawsuits against OpenAI that AI Chat Daily covered earlier this month — the judge made clear she did not want testimony about specific AI scandals. xAI and OpenAI's safety frameworks, she said, were on the table. Specific product failures were not.
Thursday's witness list runs deep. Musk returns for another round of adversarial questioning, followed by his family office manager Jared Birchall, AI safety researcher Stuart Russell, and OpenAI president Greg Brockman. Russell's appearance signals that Musk's team intends to put OpenAI's safety culture on trial, even if the judge's evidentiary rulings keep narrowing the aperture.
The trial is becoming a stress test of whether a litigant whose primary medium is X can survive a courtroom that treats every post as evidence. Musk's $100M-versus-$38M discrepancy and the AGI contradiction are not just embarrassments — they hand OpenAI a credibility argument that Savitt is clearly intent on compounding. For a case that ultimately turns on what Musk understood and agreed to between 2015 and 2018, the gap between his public claims and his under-oath answers is the entire ballgame.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
