The discovery process alone could reveal more about how AI frontier models get built than the industry has disclosed in five years.
The Summary
- Jury selection begins Monday in Oakland for Musk's $134 billion lawsuit claiming Altman betrayed his $38M early OpenAI investment by converting the nonprofit to for-profit.
- Witness list reads like a who's who of AI power: Satya Nadella, Mira Murati, Greg Brockman, and Shivon Zilis will all testify under oath.
- The case centers on whether Altman deceived Musk about nonprofit commitments, but the real show is what gets exposed in discovery about OpenAI's internal operations.
- Legal theories have shifted from breach of contract to unfair business practices to false advertising, signaling this is more vendetta than clear-cut legal strategy.
The Signal
Musk claims Altman promised his $38 million would build AI for the public good, then flipped OpenAI into a for-profit subsidiary of the nonprofit shell. OpenAI's defense: no false promises were made, the investment carried no such restrictions, and the core nonprofit still exists. Nine jurors in Oakland will decide who's lying.
But the legal outcome matters less than what gets aired in the process. Discovery in this case could expose OpenAI's biggest internal secrets, from early partnership terms with Microsoft to decision-making around GPT model releases to the actual governance structure that let a nonprofit birth a company now valued in the hundreds of billions. We've never had this level of forced transparency into a frontier AI lab.
"Questions about Altman's honesty have dogged him for years."
The witness list alone tells you this will get personal. Mira Murati, OpenAI's former CTO who briefly replaced Altman during his 2023 board coup, will testify. So will Satya Nadella, whose Microsoft has $13 billion invested in OpenAI. And Shivon Zilis, Neuralink executive and mother of three of Musk's children, who was there at OpenAI's founding. Under oath, in front of a jury.
Context matters here. Musk cofounded OpenAI in 2015, then left in 2018 when he wasn't made CEO. He's been publicly attacking the company since early 2023, which prompted Altman to text him: "it really fucking hurts when you publicly attack openai". That text will likely be exhibit A in OpenAI's argument that this lawsuit is just sour grapes from a spurned founder whose own AI company, xAI, lags behind.
Key details that will matter:
- Musk's legal theories have shifted repeatedly, suggesting weak underlying claims
- The $134 billion damages figure suggests Musk is valuing his early contribution as foundational to OpenAI's entire current worth
- OpenAI maintains it remains a nonprofit with a for-profit subsidiary, a structure that's legally novel and untested at this scale
The trial comes at a moment when AI governance questions are white-hot. How do you keep a nonprofit mission intact when the technology becomes worth hundreds of billions? Who gets to make decisions about deploying systems that could reshape human work, creativity, and knowledge? Can any individual, even one as capable as Altman, be trusted with that much power?
The Implication
Watch what comes out in discovery more than the verdict. Emails between Altman and Microsoft about the for-profit conversion. Internal memos about GPT-4's capabilities before release. Board meeting minutes from the weeks before Altman's brief firing in 2023. This trial will force into the public record information that frontier AI labs have successfully kept private for years.
For anyone building AI agents or trying to understand how the sausage gets made at the companies shaping Web4, this is required viewing. The organizational structure, the investor relationships, the internal power dynamics, all of it will be on display. And if the case exposes serious governance failures at OpenAI, expect regulators and competitors to take notes.