The courtroom discovery process is about to do what a decade of tech journalism couldn't: force OpenAI to show its receipts.

The Summary

The Signal

The evidence drop tells a founding story very different from OpenAI's polished origin myth. Court exhibits show Musk didn't just write checks in 2015. He drafted the mission statement, heavily influenced early structure, and appeared to expect significant control. Meanwhile, Greg Brockman and Ilya Sutskever were already nervous about that control before the company even had a name. The documents reveal Altman leaning on Y Combinator connections for early support and Jensen Huang quietly providing compute infrastructure that would prove critical.

This matters because it punctures the narrative of a clean ideological split. The story has always been: noble nonprofit corrupted by profit motive versus paranoid founder who couldn't handle collaborative governance. The exhibits suggest something messier: competing visions from day one, with Musk expecting operational control in exchange for capital, and Altman building a structure that would eventually price him out.

"Questions about Altman's honesty have dogged him for years. His trustworthiness is set to be front and center at trial."

The timing is deliberate torture. OpenAI is preparing an IPO that values the company north of $850 billion. You cannot take a company public while its CEO is on the stand defending whether he defrauded a cofounder who contributed tens of millions in seed capital. The discovery process will force OpenAI to produce:

  • Internal communications about the nonprofit-to-for-profit transition
  • Board meeting minutes from the critical 2018-2019 period
  • Financial projections that justified the Microsoft deal structure
  • Early governance documents that may or may not support Musk's claims about control rights

Musk boosted a Ronan Farrow investigation questioning Altman's trustworthiness the same day opening arguments began. The New Yorker profile, originally published three weeks earlier, suddenly appeared in X feeds with a label reading "This organic post was boosted by @elonmusk." This is pre-trial publicity masquerading as platform curation. The judge noticed, lecturing both parties about social media conduct.

The witness list reads like a who's who of AI's power structure. Microsoft CEO Satya Nadella, former OpenAI CTO Mira Murati, Shivon Zilis (mother of three of Musk's children and Neuralink executive), Brockman, Altman, and Musk himself. Each has incentive to tell a version of history that protects their current position.

Altman showed up for jury selection wearing a dark suit, unexpected for a defendant not required to attend unless testifying. Musk did not attend day one. Several potential jurors expressed negative views of both AI and Musk, creating an interesting baseline: a jury skeptical of the technology both men are building and at least one of the personalities involved.

The Implication

If you're building anything in the agent economy, watch what happens when founding documents meet actual incentives. The OpenAI structure was supposed to solve the alignment problem at the corporate level: a nonprofit parent controlling a for-profit subsidiary, ensuring mission over margin. That structure is now exhibit A in a fraud case. The discovery will show how easily "benefit humanity" language bends when Microsoft offers $10 billion and a path to $850 billion valuations.

For anyone structuring Web4 companies or tokenized governance, this trial is a free masterclass in what breaks when growth outpaces governance design. Musk's core claim is simple: I gave you money for X, you built Y, and that's fraud. If he wins, every AI company with "safety" or "public benefit" in its founding documents just inherited a legal liability. If he loses, it confirms what many already suspect: mission statements are marketing and control follows capital, regardless of what the slide deck said in 2015.

Sources

The Verge AI | Bloomberg Tech | Business Insider Tech | Wired AI | MIT Tech Review AI