Musk Blew Up His Own Case on the Stand. The Courtroom Gasped.

The man suing OpenAI for betrayal confirmed in a federal courtroom on May 4 that his own AI company trains on OpenAI’s models. Not alleged, not leaked — Musk said it under examination, with reporters in the room. MIT Technology Review’s in-room account describes audible gasps. The lawsuit’s moral foundation didn’t survive its first week.

The pretrial case was always structured around one clean argument: OpenAI executed a bait-and-switch, transforming a nonprofit lab into a for-profit corporation worth hundreds of billions while the founders who believed in the mission got cut out. When Musk took the stand on May 1, he called himself “literally a fool” for funding the organization. He repeated “you can’t just steal a charity” with the rhythm of someone who had rehearsed it. The argument had shape. Then came May 4, and Musk confirmed that xAI uses OpenAI’s models to train Grok. According to MIT Technology Review’s full week-one recap, this came out under examination. The man suing OpenAI for betraying the AI mission has been building his competing product on that mission’s output. The courtroom gasped because it understood immediately what that admission meant for everything Musk had spent the previous three days saying.

There’s more. CNBC reported that Musk reached out to Greg Brockman seeking a settlement two days before the trial opened. Two days. If this case is principled — if it’s genuinely about protecting the idea of AI developed for humanity’s benefit — you don’t attempt to kill it on a Wednesday when opening arguments begin Friday. The settlement attempt suggests that Musk either anticipated what the stand would expose, or that principle was always secondary to outcome. Neither interpretation flatters the plaintiff.

The counterargument worth taking seriously is that the underlying complaint has real substance, even if Musk has disqualified himself from making it. OpenAI did pivot from nonprofit to capped-profit to full for-profit in a way that altered the terms under which donors and early backers contributed. OpenAI’s own lawyer — previously Musk’s attorney at Tesla — acknowledged during proceedings that Musk was “never committed to it being a nonprofit,” which is a strange concession from the defense and simultaneously weakens every moral claim Musk is making. The trial is surfacing a genuine question: is Silicon Valley’s mission-first framing a binding legal commitment or favorable brand positioning? That question deserves serious scrutiny. Musk just isn’t the one to demand the answer.

What’s actually hanging over all of this is OpenAI’s IPO. The company is targeting a valuation around $852 billion on its way to a public listing that would be the most closely watched tech offering since Meta. If Musk’s “bait and switch” claims hold — if the court agrees that OpenAI materially defrauded its founding donors when it restructured — the legal overhang forces a restructuring that spooks institutional investors and can derail the timeline entirely. This isn’t a charity dispute being adjudicated in Oakland. It’s a multi-week proceeding with a direct line to whether the most anticipated IPO in tech history happens this decade.

Week two starts with Greg Brockman expected to testify. Sam Altman hasn’t taken the stand yet. What Musk accomplished in week one was to hand opposing counsel the best piece of evidence they’ll see all trial: his own voice, in court, confirming that the mission he came to defend is one he exploits every day. He called himself a fool in an Oakland courtroom. The jury is still deliberating on whether he was right.

Sources: MIT Technology Review · CNBC · The Ringer