新闻

A nine-person Oakland jury must first resolve whether Musk has the legal right to sue at all -- a threshold most charitable-trust scholars predict he will lose -- before reaching the $134 billion disgorgement claim, Altman's ouster, and the fate of safety obligations protecting 700 million ChatGPT users A nine-person jury in Oakland began deliberating Monday in the most consequential corporate governance trial in the history of American artificial intelligence, weighing whether OpenAI CEO Sam Altman and president Greg Brockman violated charitable-trust law when they converted a nonprofit AI research lab into an $852 billion public benefit corporation -- and, in doing so, whether the safety obligations that currently bind that company to protect its users would survive a ruling in Musk's favor or quietly dissolve in a fully commercial restructure. The answer to that question matters to every one of the more than 700 million people who use ChatGPT each week. A ruling for Musk that forces the unwinding of OpenAI's October 2025 recapitalization would introduce immediate uncertainty over the company's governance, its ability to raise capital for a projected IPO near a $1 trillion valuation, and the enforceability of the safety conditions California Attorney General Rob Bonta extracted from the company as a condition of approving that restructuring. A ruling for OpenAI, by contrast, would set a precedent making it far harder for any future donor to challenge a nonprofit AI lab's conversion to commercial status. Before the jury reaches the substance of Musk's claims -- that Altman and Brockman breached the charitable trust he established with approximately $44 million in donations between 2015 and 2017, and that both men unjustly enriched themselves through OpenAI's commercial transformation -- it must first resolve a threshold standing question: did Musk file his lawsuit within the statutory time limit? Musk departed OpenAI's board in 2018. He did not file suit until 2024 -- a six-year gap his own lawyers acknowledge they must explain. Judge Yvonne Gonzalez Rogers ruled that the jury's finding on this point would be advisory, but said she would very likely follow its recommendation. Most charitable-trust legal scholars are skeptical that the standing question survives. Under American charitable-trust doctrine, the power to police donor intent typically rests with state attorneys general, not with individual donors. The California attorney general declined to join Musk's suit and said the action did not serve the public interest. If the jury finds the claims time-barred, the case ends without reaching the merits and OpenAI's restructuring stands. If the case survives the standing hurdle, the jury will weigh two civil claims: breach of charitable trust and unjust enrichment. The most damaging piece of evidence for OpenAI is not an email from Altman. It is a personal journal entry by Brockman, written in November 2017 after a meeting with co-founder Ilya Sutskever. The entry reads: "Cannot say we are committed to the non-profit... if three months later we're doing b-corp then it was a lie." Judge Gonzalez Rogers cited that entry directly in her January 2026 ruling that allowed the case to proceed to trial, finding what she described as "ample evidence" and rejecting nearly every OpenAI motion to dismiss. Three weeks of testimony produced a parade of Silicon Valley figures -- Altman, Brockman, Microsoft CEO Satya Nadella, former OpenAI chief scientist Ilya Sutskever, and Musk himself, who testified over three days -- and reduced to two competing summary readings: that Altman and Brockman "stole a charity," as Musk's lead counsel Steven Molo told the jury, or that Musk "didn't get his way at OpenAI," as OpenAI attorney William Savitt framed it. Savitt also told jurors that Musk was not present for closing arguments -- he was in Beijing as part of President Trump's state-visit delegation -- and used the absence pointedly: "Mr. Musk isn't here today -- my clients are here. They're here because they care about this." Musk is asking Judge Gonzalez Rogers to remove Altman and Brockman from their leadership roles, unwind the October 2025 recapitalization that converted OpenAI into a public benefit corporation with Microsoft holding a 27 percent stake, and direct up to $134 billion in disgorgement from OpenAI and Microsoft to OpenAI's nonprofit foundation. On the stand, Musk renounced any personal financial benefit from the case, framing the relief as a restoration of the charitable mission he said Altman and Brockman betrayed. "If we make it okay to loot a charity," Musk told the jury, "the entire foundation of charitable giving in America will be destroyed." The $134 billion figure, however, faces a significant obstacle independent of the liability question. At a March 2026 pretrial hearing, Gonzalez Rogers expressed open skepticism about the damages theory, calling it unconvincing. OpenAI has projected losses of $14 billion in 2026. The simultaneous remedies phase that opened Monday -- heard by Gonzalez Rogers alone, without jury input -- will determine what penalties, if any, follow a liability finding. OpenAI's attorneys offered a competing account. Sarah Eddy told the jury that Altman and Brockman never made legally binding commitments to Musk about corporate structure, and that his donations carried no enforceable conditions. She pointed to evidence in the court record showing that Musk himself had pushed, as early as 2017, to create a for-profit subsidiary -- on the condition that he control it. When that demand was refused, he left the board in 2018. Four years later, he founded his own AI company, xAI. A year after that, he filed suit. "He wanted dominion over AGI," Eddy told the jury. "That's why this was such a high-stakes conversation." Altman's own testimony described Musk's 2018 departure as both disruptive and, for many employees, a relief. He testified that Musk had proposed the company eventually pass to his children upon his death -- a prospect Altman said he found alarming. Microsoft attorney Russell Cohen added that the tech giant had been "a responsible partner at every step" and had no knowledge of conditions Musk may have placed on his charitable contributions. The trial drew testimony that cut against both parties. Ilya Sutskever -- who co-founded OpenAI and played a central role in Altman's brief ouster from the CEO position in November 2023 -- testified that he had spent months gathering evidence of what he called Altman's pattern of deception before voting to remove him, then later reversed that vote and said he regretted reinstating Altman. Altman, under cross-examination from Musk's attorney Steven Molo, was asked: "Are you completely trustworthy?" He answered that he believed he was, though he acknowledged he had "told the occasional lie." Molo later told jurors that five witnesses -- including Musk, former board members, and Sutskever -- had described Altman as dishonest. Former OpenAI board member Shivon Zilis, the mother of four of Musk's children, was called by OpenAI to testify. Her testimony did not corroborate Musk's account of the founding commitments. Eddy used that fact in closing, telling the jury that "even the mother of his children" could not back Musk's version of events. The nonprofit obligations underlying this trial are not abstract. California Attorney General Bonta's October 2025 agreement with OpenAI requires the company to mitigate risks to teenagers and maintain AI safety oversight provisions -- obligations that exist specifically because the nonprofit structure survived the recapitalization. A complete unwinding of that structure, as Musk requests, would not automatically preserve those conditions. An advisory verdict is not a binding one, and the remedies available to Gonzalez Rogers are constrained by law, but the governance architecture this case builds around charitable-trust enforcement will determine whether future AI labs converting from nonprofit status face any meaningful legal accountability at all. That is why the case has been described by both sides, with rare agreement, as one that will define AI governance for a decade. In April 2025, twelve former OpenAI employees -- represented by Harvard law professor Lawrence Lessig -- filed an amicus brief stating that the company had abandoned its nonprofit roots and that Altman was "a person of low integrity who had directly lied to employees." Rose Chan Loui, founding executive director of the Lowell Milken Center on Philanthropy and Nonprofits at UCLA Law School, said the conversion amounts to "a change of purpose" requiring court approval. Anthropic -- which, like OpenAI, began with a mission-driven structure -- has watched the proceedings closely. So have AI governance regulators around the world. An advisory verdict for Musk could arrive as early as this week. It would not, by itself, remove Altman from his job -- that determination rests with Gonzalez Rogers in the remedies phase. It would not automatically unwind Microsoft's $13 billion investment or freeze the IPO path. What it would establish -- or foreclose -- is whether the courts are prepared to treat nonprofit-to-profit conversions in AI as a category of corporate conduct subject to legal challenge by past donors and the public they originally served. If the jury finds for OpenAI or rules the claims time-barred, the company's current trajectory continues: a public benefit corporation valued between $500 billion and $1 trillion, with a nonprofit foundation holding approximately 26 percent, preparing for a public market debut. If the jury finds for Musk and Gonzalez Rogers follows, the legal architecture OpenAI built during the most expensive AI arms race in history faces a court-ordered unwinding -- and the question of who governs the world's most widely used AI platform returns to a judge in Oakland.

The next wave of pre-IPO perp markets on HIP-3 includes SpaceX, Anthropic, and OpenAI IPOs, all of which are targeting $1 trillion-plus valuations. Hyperliquid's native token HYPE is outperforming the broader crypto market slump as its pre-IPO perpetual futures ecosystem gains momentum ahead of a historic wave of technology listings. HYPE is trading at $45.17 on Monday, up roughly 5% over the past 24 hours and 69% over the past year, according to CoinGecko data. The divergence is tied to Hyperliquid's HIP-3 marketplace, where traders are betting on pre-IPO shares of SpaceX, Anthropic, and OpenAI, a $120 billion volume opportunity that traditional finance cannot offer. The HIP-3 framework, which allows third-party teams to launch their own perpetual futures markets, has processed more than $120 billion in total volume since launch, according to Dune Analytics data. On April 8, HIP-3 deployers generated 48.1% of Hyperliquid's total platform volume, approaching parity with the platform's own native markets. That parity means retail traders can now access pre-IPO price exposure that was previously available only to institutional investors through secondary venues. TradeXYZ, the leading HIP-3 deployer, demonstrated the model works when it priced Cerebras perpetuals within 3% of the AI chipmaker's Nasdaq debut while traditional secondary platforms were 35% off. The gap showed that on-chain pre-IPO markets achieved faster, more accurate price discovery than off-chain alternatives. "For years, retail investors often entered once companies were already public and much of the upside had already played out," Diego Martin, CEO of Yellow Capital, told Decrypt. "This feels like the beginning of a much bigger shift in who gets to participate." The next cohort of test cases is already forming, with multiple popular IPOs lined up for 2026. SpaceX is targeting a June IPO that could raise between $75 and $80 billion, roughly double the total raised across all 2025 IPOs combined, at a valuation of up to $1.75 trillion to $2 trillion, according to Reuters. Anthropic and OpenAI are each eyeing listings that could raise $60 billion at valuations exceeding $1 trillion. A SpaceX pre-IPO perpetual contract, SPCX, is already live on TradeXYZ with a $150 reference price. That pricing implies a valuation of roughly $1.78 trillion. The token is currently trading at $207, down roughly 10% from Monday's $230 local top. Market expectations are already pricing in the magnitude of these listings. On prediction market Myriad, owned by Decrypt's parent company Dastan, users assign a 91% chance that SpaceX's closing market cap will exceed $1.3 trillion, and give Anthropic a 67% chance of going public before OpenAI. Matthew Pinnock, COO of Altura DeFi, told Decrypt the Cerebras market showed how on-chain infrastructure can outpace traditional secondary markets on price formation. "24/7 crypto rails, leveraged positioning, and global participation can create faster consensus formation around late-stage private assets, particularly for sectors like AI where demand is heavily retail-driven and information moves quickly," he said. The category carries meaningful regulatory risk, Pinnock added, and he expects regulators to eventually scrutinize whether pre-IPO perpetual products function as unregistered securities exposure for retail traders. OpenAI and Anthropic have already warned investors against trading in securities tied to companies that they have not authorized. The platform's ambitions have drawn early attention from Wall Street: Intercontinental Exchange and CME Group have reportedly urged the CFTC to address potential market integrity risks associated with Hyperliquid's pseudonymous trading environment. The Hyperliquid Policy Center pushed back, arguing the platform's transparency is "hostile" to insider trading. "Instead of being focused purely on digital assets, crypto infrastructure starts becoming a way for people to participate in broader financial opportunities," Martin said. "That is a story a much larger audience can understand."