There’s a sentence buried in the trial coverage this week that deserves more attention than it’s getting.

Sam Altman testified that Elon Musk, during OpenAI’s early days, mulled handing the organization to his children. That’s it. Sit with that for a second. OpenAI was founded on a single animating idea: that advanced AI is too dangerous to end up in the hands of any one person. And the co-founder pushing hardest for control of the thing was apparently thinking about keeping it in the family.

You genuinely cannot make this stuff up.

What the trial is actually about

Musk’s lawsuit against OpenAI has been dressed up in the language of broken promises and nonprofit betrayal. His argument, roughly, is that OpenAI sold him a vision of open, safety-focused AI research and then converted into a for-profit company once the money got serious. He wants the conversion blocked or reversed.

There’s a version of that argument that’s serious and worth engaging with. The original OpenAI charter was an unusual document. It committed the organization to developing AI for the benefit of humanity broadly, not shareholders. The move toward a capped-profit structure, and the deepening relationship with Microsoft, genuinely did change what OpenAI is. Reasonable people can debate whether those changes violated the spirit of the founding.

But then you get into the testimony, and the serious argument falls apart pretty fast.

Altman told the jury that Musk’s focus during the early days wasn’t on the mission. It was on control. Specifically, Musk reportedly pushed to control the initial for-profit structure of the company. Altman said this gave him pause, because OpenAI was built on the premise that keeping advanced AI out of any single person’s hands was the whole point. And Altman, from his years running Y Combinator, knew that “founders who had control usually did not give it up.”

That’s a polite way of saying: I knew exactly what Musk was going to do with it.

The chainsaw

There’s another detail from Altman’s testimony that I keep coming back to. Musk, according to Altman, required OpenAI president Greg Brockman and former chief scientist Ilya Sutskever to rank researchers by their accomplishments and then “take a chainsaw through a bunch” of them.

Altman’s framing was that this was just Musk being Musk, that it’s the management style he’s known for, but that it was incompatible with OpenAI. Which is true, as far as it goes. But there’s something more interesting underneath it.

AI research, particularly safety research, is not a factory floor. You can’t run it like Tesla production. The researchers who do the most important work are often the ones working on problems that don’t have obvious near-term payoffs, on questions that make the models slower and more constrained rather than faster and more impressive. Running a chainsaw through your research org, sorted by “accomplishments,” is a good way to systematically eliminate the people thinking hardest about whether any of this is actually a good idea.

Musk’s stated reason for co-founding OpenAI was that he was worried about AI safety. His actual management instincts, at least as Altman describes them, would have gutted the safety-focused researchers first.

Altman on the stand

Altman himself was, by all accounts, good on the stand. After two weeks of witnesses describing him in unflattering terms, he showed up in full “reasonable, slightly bashful tech founder” mode and delivered. His closing line when his own lawyer asked him how it felt to be accused of stealing a charity: “We created, through a ton of hard work, this extremely large charity, and I agree you can’t steal it. Mr. Musk did try to kill it, I guess. Twice.”

That’s a great line. Crisp, a little wounded, a little wry. The jury probably liked it.

But I’d note that Altman winning on the stand isn’t the same thing as OpenAI being vindicated. The discomfort at the heart of this trial isn’t really about Musk, or at least not only about Musk. It’s about what happens when you try to build a safety-focused nonprofit and then discover that the thing you built is worth hundreds of billions of dollars.

Musk’s alleged bad behavior during the founding era is real, if the testimony is accurate. But OpenAI also did restructure in ways that moved it away from its original model. Both things can be true. Altman being a better witness than Musk doesn’t settle the underlying question of whether the nonprofit mission survived the money.

The inheritance angle

Back to the children thing, because I don’t want to let it go.

The OpenAI charter is explicit: “We are concerned about late-stage capitalism, of AI being used to unduly concentrate power.” One of the explicit failure modes the organization says it’s trying to prevent is any group, including OpenAI itself, seizing “unprecedented societal control.”

What’s a more textbook example of “unprecedented societal control” than one billionaire’s children inheriting the organization developing the most powerful AI systems in the world?

I’m not saying Musk was seriously planning this. It may have been an offhand comment. Altman’s testimony filtered through lawyers and memory isn’t a transcript. But the fact that this was apparently in the conversation, even as a passing thought, tells you something about the gap between the mission as written and the mission as actually understood by the people in the room.

Why it matters beyond the courtroom

The Musk v. OpenAI trial is, on one level, a weird celebrity grudge match dressed up in legal briefs. It’s Elon being Elon, nursing a grievance, spending money to litigate a relationship that soured a decade ago.

But the questions underneath it are ones the whole industry is going to have to keep answering. Who actually controls these organizations? What does a “safety mission” mean when the organization pursuing it is worth more than most countries’ GDP? When the founder who wanted AI kept out of any single person’s hands apparently wanted to keep it in his own family, what does that tell us about how much any of these stated missions should be trusted?

Altman probably wins this trial. His performance on the stand was strong, the facts seem to favor him, and Musk’s legal theory has always been a stretch.

But winning a lawsuit about whether you honored your founding mission isn’t the same as having honored your founding mission. The trial has been useful precisely because it’s forced into public record what “safety-first AI development” actually looked like when powerful people were fighting over who got to run it.

It looked, often, a lot like everything else.