Musk-Altman Trial Surfaces New Allegations on OpenAI's Governance and Founding Disputes

Three weeks of testimony in a federal courtroom in Oakland, California have laid bare the messy origins of the most consequential AI company in the world. The trial of Musk v. Altman — a $134 billion lawsuit alleging that OpenAI's leadership "stole a charity" — concluded closing arguments on May 14, 2026, with a nine-person jury set to begin deliberations [1][2]. What emerged was not just a contract dispute between billionaires but a window into how Silicon Valley's most powerful figures make and break promises when world-changing technology is at stake.

The Claims at the Heart of the Case

Of the 26 claims Elon Musk originally brought against OpenAI, CEO Sam Altman, and President Greg Brockman, Judge Yvonne Gonzalez Rogers dismissed all but two before trial: breach of charitable trust and unjust enrichment [3]. The surviving claims center on whether Altman and Brockman violated their fiduciary duties to OpenAI's nonprofit mission when the company shifted to a capped-profit structure in 2019 and later pursued full for-profit conversion.

Musk's legal team, led by attorney Steven Molo, argued that Altman systematically maneuvered to convert charitable assets — built with donor contributions — into personal wealth. They pointed to OpenAI's founding charter from 2015, entered into evidence, which declared the organization would pursue "open source technology for the public benefit" and was "not organized for the private gain of any person" [4]. Incorporation documents showed a governance plan envisioning a five-person board, originally including Musk and Bill Gates, to decide on AGI's application "for the good of the world" [4].

OpenAI's defense, meanwhile, characterized Musk as a disgruntled former donor who abandoned the organization and now seeks to cripple a competitor. Altman testified that the nonprofit was "left for dead" after Musk withdrew support, and that the shift to a profit structure was essential to attract the capital needed to pursue their mission [5].

The Money Gap: What Musk Actually Contributed

One of the trial's most concrete revelations concerned the gap between what Musk pledged and what he actually gave. When OpenAI launched in 2015, a total of $1 billion was pledged by Musk, Altman, Greg Brockman, Reid Hoffman, Jessica Livingston, Peter Thiel, Amazon Web Services, and Infosys [6]. But actual collections through 2021 totaled only $133.2 million [6].

Source: TechCrunch / Court Filings

Data as of May 1, 2026CSV

Musk's individual contribution has been a moving target. He publicly claimed $100 million in earlier statements, then revised it to $50 million [7]. OpenAI maintains he donated $38 million [8]. Only approximately $15 million can be definitively traced to Musk through public filings [7]. During trial testimony, Musk acknowledged he "didn't come close" to the $1 billion he pledged, accusing OpenAI of "looting the nonprofit" [9].

The shortfall matters legally. If Musk contributed $38 million out of a $133 million total, he provided roughly 29% of the organization's actual funding — substantial, but far less than the controlling stake his original pledge implied. OpenAI's lawyers argued this undermines his standing to claim breach of founding intent: he didn't fund the organization at the level he promised, then left, and now wants to control what others built [5].

For comparison, Reid Hoffman contributed through Silicon Valley Community Foundation and other vehicles; combined non-Musk contributions totaled approximately $75.8 million [7]. The other co-founders collectively outspent Musk.

Five Witnesses Called Altman a Liar

The most damaging testimony against Altman came not from Musk but from former colleagues. During closing arguments, Molo cited five witnesses who described the OpenAI CEO as dishonest: Musk himself, co-founder and former chief scientist Ilya Sutskever, former CTO Mira Murati, and former board members Helen Toner and Tasha McCauley [10][11].

Sutskever testified that he spent months gathering evidence showing what he described as "Altman's pattern of deception and poor management" [10]. Toner and McCauley — who did not appear in person but whose edited deposition tapes were played for jurors — described "a toxic culture of lying" at OpenAI and said Altman had not been candid with them about major decisions [11][12].

A specific example: when OpenAI released ChatGPT in November 2022, the board was not informed in advance and learned about it from Twitter [12]. McCauley confirmed this in her deposition. Murati introduced evidence of what she characterized as "leadership deceptions" and Altman's resistance to board oversight [10].

Altman responded on the stand that he believed he was "an honest and trustworthy business person," while acknowledging "there had been times he had not told the full truth" [13]. His defense framed these as communication failures, not fraud.

The November 2023 Firing: Board Authority and Evidence

The four board members who voted to fire Altman on November 17, 2023 were Toner, McCauley, Sutskever, and Adam D'Angelo [14]. Their stated rationale was that "the board no longer has confidence in his ability to continue leading OpenAI." Trial testimony revealed that the firing was not impulsive — the board had been "watching and documenting" concerns for at least a year [12].

Evidence entered at trial showed the board had documented instances of Altman withholding information, including about his ownership of OpenAI's startup fund and the ChatGPT release [12]. However, the rapid reinstatement of Altman — within five days, after a near-mutiny by employees — raises questions about whether the board's legal authority to fire him without triggering obligations to capped-profit investors was properly exercised.

The governance documents showed that under OpenAI's structure at the time, the nonprofit board had sole authority to hire and fire the CEO without needing investor approval. But the chaotic aftermath suggests the practical power had already shifted to the for-profit subsidiary's stakeholders — Microsoft, employees with equity, and investors who threatened to leave [14].

The For-Profit Conversion: Legal Scrutiny

OpenAI's 2019 conversion from a pure nonprofit to a "capped-profit" entity — and its subsequent push toward full for-profit status — has drawn serious regulatory attention independent of the Musk lawsuit.

California Attorney General Rob Bonta launched an investigation into the conversion in early 2025 [15]. A coalition of public interest nonprofits and labor advocates filed an administrative petition in April 2025 asking Bonta to halt the transition [16]. The San Francisco Foundation and over 50 organizations formally requested legal action to "protect OpenAI's charitable assets," with estimates of at-risk value as high as $157 billion [17].

Under the restructuring deal eventually struck with Bonta's office, the newly formed OpenAI Foundation would hold approximately 26% of OpenAI's valuation — about $130 billion — making it one of the wealthiest philanthropic organizations in the world [18]. But critics argue the structure is hollow: the for-profit entity could end up "calling the shots" despite the nonprofit's nominal control, and seven of the nonprofit's eight board members also sit on the for-profit's board [18][1].

Public Citizen wrote to both Bonta and Delaware's Attorney General urging rejection of the plan, calling the $100 billion stake arrangement insufficient to protect the charitable purpose [19]. No state attorney general has filed suit to block the conversion outright, but the investigation remains active and the trial's outcome could force the issue.

What's New vs. What Was Already Known

Musk's original February 2024 lawsuit contained several allegations that were re-litigated at trial: that OpenAI broke its founding commitment to open-source AI research, that the Microsoft partnership compromised independence, and that the for-profit conversion violated charitable purpose [3].

Genuinely new revelations at trial included:

• Altman's Helion conflicts: Altman confirmed under cross-examination that he owns roughly one-third of Helion Energy, a nuclear fusion company valued at approximately $5 billion in late 2025, and that he tried to steer OpenAI to purchase power from Helion [20]. The U.S. House oversight committee subsequently launched an investigation into these potential conflicts [20].
• The Shivon Zilis testimony: Musk became visibly flustered when questioned about his relationship with Zilis, an OpenAI board member who is the mother of four of his children. Zilis testified that Musk had offered her his sperm after noticing she had no children while she served on the board [21].
• Free Teslas for influence: Evidence included communications about free Tesla vehicles allegedly offered in exchange for influence over OpenAI decisions [21].
• The 2017 governance split: Internal emails from 2017-2018 showed key leaders demanding an "ironclad agreement" to prevent any single person from controlling AGI, specifically rejecting Musk's push for "less time and more control" over the lab [4].

The timing question looms large. Musk founded xAI in 2023 as a direct OpenAI competitor, and recently merged it with SpaceX in a deal valuing the combined entity at $1.25 trillion [3]. OpenAI's legal team repeatedly characterized the lawsuit as competitive sabotage timed to xAI's launch.

Governance Structures Compared

OpenAI's hybrid model — a nonprofit board nominally controlling a for-profit subsidiary — is one of several experiments in AI governance. Each has distinct failure modes:

OpenAI (Nonprofit → Public Benefit Corporation): The nonprofit retains the ability to appoint and remove all directors of the for-profit entity. But as the November 2023 crisis showed, practical power often lies with the stakeholders who control capital and talent. The Harvard Law Review has described this as "amoral drift" — the gradual displacement of mission by market incentives [22].

Anthropic (Public Benefit Corporation with Long-Term Benefit Trust): Anthropic created a Long-Term Benefit Trust that gives power over the company's direction to trustees charged with representing the public interest [23]. This is arguably the most formalized internal governance approach among major AI labs, embedding accountability from incorporation. Its weakness: the trust's enforcement mechanisms have not been tested under pressure.

DeepMind (Alphabet subsidiary): Acquired by Google in 2014 on the condition that an independent ethics board would oversee AGI development [23]. That board's actual authority within Alphabet's corporate structure remains opaque, and DeepMind's 2018 AI Principles are non-binding commitments.

Legal scholars at Lawfare have noted that in all these structures, "no one can effectively enforce social purpose" because "corporate boards face powerful pressure from investors to deviate from social goals and yield to profit maximization" [24].

Financial Exposure and the Funding Cliff

Source: Crunchbase / Public Filings

Data as of May 1, 2026CSV

If Musk prevails on even one claim, the consequences for OpenAI's capital structure could be severe. The lawsuit seeks to reverse the public benefit corporation conversion, remove Altman and Brockman from leadership, force disgorgement of compensation and equity granted during the transition, and impose limits on OpenAI's ability to license technology exclusively to Microsoft [3].

SoftBank's $40 billion commitment — OpenAI's largest single funding round — was structured assuming continuity of the public benefit corporation [3]. An injunction halting the conversion would trigger uncertainty across that capital stack. Microsoft has booked tens of billions in projected revenue from the OpenAI partnership. Even Nvidia faces indirect exposure as one of OpenAI's largest hardware suppliers [3].

Whatever the trial court rules, the loser will appeal to the Ninth Circuit, where the case is expected to be heard in late 2026 or early 2027 [3]. The prolonged uncertainty itself is damaging — OpenAI's planned IPO timeline and employee equity values hang in the balance.

The Competitive Harm Defense

OpenAI's strongest counterargument is that the lawsuit itself is the weapon. Musk runs a direct competitor. The trial consumed three weeks of Altman's time and executive bandwidth. The overhang of a potential $134 billion judgment creates financing uncertainty that benefits xAI [3].

Notably, Musk traveled to China during the trial's final week, prompting his own lawyer to apologize to the jury for his absence [25]. The optics undercut the narrative of a wronged donor seeking justice.

Judge Gonzalez Rogers has not publicly signaled skepticism of Musk's standing, having allowed two claims to proceed to trial. But the earlier dismissal of RICO, antitrust, and breach of contract claims suggests the court found much of Musk's theory legally insufficient [3]. For a court to sanction the lawsuit as abusive litigation, OpenAI would need to demonstrate both that the claims lack merit and that they were filed primarily for competitive advantage — a high bar under federal rules.

Musk has pledged to donate any trial winnings to the OpenAI nonprofit [26], a move his supporters frame as evidence of good faith and his critics call a public relations gambit that costs him nothing unless he wins.

What Comes Next

The jury's verdict is advisory — Judge Gonzalez Rogers will make the final ruling on the equitable claims [2]. Deliberations begin the week of May 19. The outcome will determine whether OpenAI's for-profit conversion stands, whether Altman and Brockman face personal liability, and whether the charitable assets built with donor money will be protected or privatized.

Beyond the courtroom, the case has already achieved something regardless of verdict: it has forced into public view the documents, communications, and testimony that reveal how AI's most powerful institution was built on a foundation of personal relationships, shifting promises, and irreconcilable visions for technology's role in society. Whether that foundation was fraudulent or merely human is now for a jury to decide.