The jury leaves us hanging

Ironies continue, even as trial ends

IT DIDN’T END WITH A BANG, BUT A WHIMPER.

After less than two hours of deliberation on Monday, a jury in Oakland rejected Elon Musk’s claims against OpenAI and its leaders — CEO Sam Altman and President Greg Brockman.

The unanimous ruling by the six-woman, three-man jury came quickly after deliberating over a trial record that contained 11 days of live testimony from more than 20 witnesses and hundreds of exhibits.

The jury found that Musk had filed his claims for breach of charitable trust and unjust enrichment too late and that they were therefore barred by the statute of limitations.

Musk’s claims against Microsoft for aiding and abetting also failed because, without the claim for breach of charitable trust, the aiding and abetting claim also failed.

The jury’s verdict was read by the court clerk in front of U.S. District Judge Yvonne Gonzalez Rogers, in the federal courthouse in Oakland where she has presided over the trial that began April 27. While the jury was technically an “advisory jury” and Gonzalez Rogers not bound to follow its determination, she immediately accepted the verdict.

The jury’s decision came at the close of the “liability phase” of the trial during which the jury evaluated Musk’s claims that the OpenAI defendants, aided and abetted by Microsoft, breached the charitable trust allegedly created by Musk’s $38 million donations to the nonprofit, and were “unjustly enriched” by doing so. Musk claimed that the defendants collectively should cough up $134 billion in ill-gotten gains.

A second phase to determine what damages, if any, resulted from the claims, was underway at the time of the ruling and will not be continued.

The case — sometimes billed as “the trial of the century” — attracted global attention, fueled by the celebrity of the controversial plaintiff, the staggering amount of money involved, and a technology predicted to cause as profound a change in modern life as the internet or the printing press.

A resounding thud

The jury’s decision landed anticlimactically.

Steven Molo, Musk’s lead lawyer, called the decision “technical.” He said that because the case was dismissed on the grounds of the statute of limitations, the jury did not determine the substance of Musk’s allegations. That means that the extensive trial testimony about whether there was a breach of charitable trust was left unresolved.

Molo said that he respected the jury’s decision but there were strong grounds for appeal. He said that it would be appropriate for an appeal to move forward without waiting for a resolution of the other claims and counterclaims that were not expedited by the judge. (This trial was about only three of Musk’s dozens of claims carved out from the others and accelerated for trial because of their importance to the public.).

Molo expects to ask the court’s approval to do that.

The ending was unsatisfying to many court-watchers, who found it ironic that of all the important issues in the case, the least interesting was whether Musk waited too long to sue.

William Savitt, OpenAI’s lead lawyer, said that his clients were happy with the decision even though it did not resolve the question about whether his clients had remained true to the mission. Savitt said the case was simply the work of a competitor who did not get his way.

The ending was unsatisfying to many court-watchers; who found it ironic that of all the important issues in the case, the least interesting was whether Musk waited too long to sue.

James Rubinowitz, a New York lawyer who teaches a course on AI in litigation at Cardozo School of Law, summed the trial up as anticlimactic in an email to a reporter. “The whole thing ended on a calendar. Not on whether Sam Altman stole a charity. Not on whether OpenAI breached its mission…. The jury never reached a single question this trial was supposed to answer.”

The ironic outcome wasn’t out of character for the case; the trial was riddled with irony. On the 11th day of trial, Molo and Savitt were arguing about whether there was a need to correct a Molo misstatement. Molo said it was ironic that the focus was on his misstatement not Altman’s alleged lying.

Gonzalez Rogers, a no-nonsense judge who doesn’t care for lawyer histrionics, cut him off, saying sharply, “Mr. Molo, this entire trial is a giant irony.”

She didn’t identify the ironies, but there were plenty of candidates, even more if you include those cousins of irony — hypocrisies.

Ironies (and hypocrisies) in the courtroom

First, there was Musk, the world’s richest person, criticizing OpenAI’s leaders for personally obtaining massive benefits from the value created by artificial intelligence.

Then there’s the irony that Musk claimed that the risks posed by artificial general intelligence (AGI) were so great that it was best pursued in a nonprofit enterprise; he said he believed it was important to remove the profit incentives from those building the technology. Yet in 2025, Musk put together a consortium of for-profit investors who tried, unsuccessfully, to buy the assets of the OpenAI nonprofit for $97 billion. Musk is now developing AGI, not in a nonprofit company, but as part of SpaceX, a for-profit company that he controls.

And there was Musk – who testified that AGI must be safely developed — calling a young OpenAI safety researcher a “jackass” at an all-hands meeting. He had cautioned Musk that a race for AGI could be unsafe.

But the ironies weren’t limited to Musk.

Greg Brockman testified that OpenAI’s mission was his number one priority, while writing notes in his journal about the best way to get a billion dollars for himself (a quandary he’s since solved 30 times over).

Irony showed through again in the testimony about the six weeks in August and September 2017 when Musk, Altman, Brockman, and Ilya Sutskever, OpenAI’s chief scientist, were discussing the proper structure for the future of OpenAI. Few would dispute that the four are among the smartest people in the country, but they weren’t smart enough to reach agreement on how to build the technology while retaining strong independent governance.

And then there is Sam Altman. Fired in 2024 for not being “consistently candid” with the governance board, he now chairs that board as well as that of OpenAI’s for-profit subsidiary. Today the nonprofit board and the for-profit board are identical, save for one person; the likelihood that the former will overrule the latter seems only a theoretical possibility.

The ironies don’t stop there.

David Schizer, one of Musk’s experts, testified that in the world of nonprofit corporations, it is customary for people to work for less than they might get in the for-profit world — they understand the mission is the goal. Yet the founders of the OpenAI nonprofit seemed to accept that in the world of Silicon Valley, top-tier talent couldn’t be obtained without for-profit level compensation. Mission alone would never be enough.

So there was irony in seeing a federal judge on a government paycheck working with ferocious intensity, every bit as focused as the attorneys who were billing their billionaire clients at stratospheric rates. The judge was not working for special pay or a “share of equity” but to pursue her mission of delivering justice. That irony repeated itself again and again in the work of the law clerks, the federal marshals, and the court reporter, not to mention the nine jurors who’ve given up weeks of their time for a modest per diem stipend.

One incident vividly illustrated the irony of Silicon Valley billionaires and millionaires spending three weeks in a courtroom where the idea of public service isn’t dismissed but lauded.

While Gonzalez Rogers kept the trial efficiently on track, there were several moments when the proceedings were delayed. To keep the jury engaged, Gonzalez Rogers pointed to the portraits of federal judges on the courtroom’s walls. She identified the individual judges and told the jurors a little about their careers. She explained that all the judges on the wall had reached retirement age; they were no longer required to keep hearing cases, but they did, even though their paychecks were exactly what they would be if they stopped working. They still worked, essentially for free.

There was even irony in one of the court’s rulings. Gonzalez Rogers refused to let Stuart Russell, perhaps the most distinguished AI safety researcher in the world, tell the jury that among the risks of AGI was the possibility of an event that would extinguish humankind. Saying that it would be prejudicial, Gonzalez Rogers wouldn’t allow the word “extinction” to be used.

Rubinowitz, the AI in litigation lecturer at Cardozo School of Law, added a further irony. He said, “Even in victory, OpenAI walks away with the worst documentary evidence about its governance now permanently in the public record. … A statute of limitations ruling does not unburn what was burned during three weeks of testimony.”

AI ironies

Ironies abounded beyond the courtroom, too. Bay City News tried an experiment with ChatGPT, the chatbot developed by OpenAI, and Grok, the bot owned by Musk’s company SpaceX.

BCN gave the bots the jury instructions, the verdict form, and a zip drive of 30-plus stories that BCN reporters have written about the lawsuit dating back to its first filing in San Francisco state court in February 2024.

BCN asked the bots to fill out the jury form in accordance with the judge’s instructions, as though the bots were jurors.

In this “trial” one of the bots ruled for Musk and against OpenAI, Altman and Brockman, finding there was a breach of charitable trust and unjust enrichment and that Musk’s suit was timely.

The other bot ruled against Musk on everything.

The pro-Musk bot? OpenAI’s ChatGPT.

The anti-Musk bot? Musk’s Grok.

Then BCN asked the bots to put aside the trial and outline what they thought would be the best resolution of the matter.

ChatGPT explains the healthiest outcome

ChatGPT said, “The best long-term resolution would not be ‘Musk wins’ or ‘Altman wins’ in a personal sense. It would be a settlement or governance reform that makes OpenAI’s public-benefit obligations more concrete, enforceable, and independently monitored, while still allowing it enough commercial capacity to fund frontier AI development.”

It added that the “healthiest outcome” would include a mission that was “not just aspirational but operational,” including separation of the oversight people from the people whose decisions they are reviewing, and transparency in the form of “stronger public reporting, outside audits, red-team review and regulatory access.”

ChatGPT cautioned that a solution should not be personality-based: “Neither founder should be treated as the embodiment of the public interest. Musk raises serious governance concerns; Altman/OpenAI raise serious institutional-capacity arguments. The answer should be structural, not charismatic.”

The bot’s bottom-line view? “The fight should be resolved by converting the dispute into enforceable governance guardrails.”

Grok

Grok’s view was similar. The x.AI bot said that it would “favor a negotiated settlement or narrow judicial ruling that validates the original safety intent without bankrupting OpenAI or handing de facto victory to any one faction.”

It said, “Musk has legitimate grievances and has driven real competition/safety awareness via x.AI. Altman/Brockman/Microsoft advanced capabilities rapidly (with real safety work alongside). Forcing a binary ‘winner’ damages the field.”

Grok went on, “The long-term path to safe AGI is messy pluralism: rival labs racing and cooperating on standards, backed by transparent evaluation and external checks.”

It concluded that, “Long-term beneficial AI is more likely from hybrid ecosystem + international norms than one lab’s internal ‘superalignment’ team. Over-reliance on any single company’s promises (whether OpenAI’s ‘superalignment’ or x.AI’s ‘understand the universe’) is naive.”

While the bots picked different winners in the litigation, they both saw the best and healthiest outcome in forward-looking guardrails and regulation of the development of AGI.

BCN then asked Grok what it thought about ChatGPT’s proposal for a resolution. Grok said, “I strongly agree with this approach … it sets a healthy precedent for how powerful AI organizations should be governed. This is exactly the kind of pragmatic, institution-focused outcome the field needs. Well said.”

When BCN reversed the question, ChatGPT wasn’t completely sold on Grok’s “messy pluralism,” but it thought that Grok’s ideas were “a sophisticated and largely persuasive framework. It correctly recognizes that the central risk is not simply ‘who wins the lawsuit,’ but how to preserve accountability, competition and safety simultaneously in a field where excessive concentration of power — whether corporate, ideological or personal — could become dangerous.”

Maybe the greatest irony is that the advanced intelligences developed by these feuding founders turn out to be better at finding agreement with each other than are the founders themselves.

The post Musk v. Altman — Day 13: Blockbuster AI trial ends without answering its biggest questions appeared first on Local News Matters.

Musk v. Altman verdict: Musk’s $134B OpenAI suit collapses after swift jury deliberations