The Journalist OpenAI Froze Out Just Got the Last Laugh

A California jury ruled on May 18 that Elon Musk waited too long to sue OpenAI. The statute of limitations had expired on his core claims. The jury found for Sam Altman and Greg Brockman. Legally, OpenAI won on a procedural technicality that has nothing to do with the merits of whether the company betrayed its founding mission.

Karen Hao spent three years finding out the answer to that question the hard way.

In early 2019, Hao became the first journalist to embed within OpenAI. Three days later, after observing a culture so secretive that a security guard had been given her photograph and employees were warned on Slack not to speak to her beyond sanctioned conversations, she published her findings in MIT Technology Review: OpenAI's leadership, despite publicly heralding a mission to build artificial general intelligence, could not actually say what that meant or how they intended to get there. OpenAI responded by refusing to speak to her for three years.

Six years later, the trial record caught up with what Hao reported privately. The emails, memos, and text messages between OpenAI's founding team — entered into the court record in the Musk v. Altman case — confirm exactly what Hao described in 2020. Leadership could not define AGI. The mission was fluid. The pivot to commercial priorities was already underway in the founding conversations themselves.

"The trial gave a tantalizingly rare window into the origins of the company," Hao told The Times. "Watching it, I felt vindicated. It was good to see lots of what I had discovered being laid out."

The statute of limitations does not adjudicate facts. The evidence that Hao gathered through three days of embed access and dozens of interviews with former employees — the evidence that earned her three years in OpenAI's cold — is now confirmed by the company's own founding documents. What the jury decided is separate from what the emails prove.

OpenAI is now preparing to file for an initial public offering. The company is working with Goldman Sachs and Morgan Stanley on a prospectus, with a target listing as early as September. The numbers tell their own story: $25 billion in annualized revenue as of February 2026, up from $21.4 billion at the end of 2025. An $852 billion valuation in its most recent funding round. A nonprofit born to save humanity that has become one of the most valuable commercial enterprises in the history of American business.

The trial record shows how early the commercial pivot was already baked in. In one email from May 2015, before OpenAI existed, Sam Altman wrote to Elon Musk: "Been thinking a lot about whether it's possible to stop humanity from developing AI. I think the answer is almost definitely not. If it's going to happen anyway, it seems like it would be good for someone other than Google to do it first." The mission framing came later, in the founding documents that Musk himself largely drafted: digital intelligence "unencumbered by an obligation to generate financial returns". That language is now in the trial record. It is also, by any honest reading, the opposite of what OpenAI has become.

Hao expanded her reporting into a book, Empire of AI: Inside the Reckless Race for Total Domination, published by Penguin Press in May 2025. The 496-page account traces the arc from nonprofit idealism to commercial engine, documenting along the way the ghost workers in Colombia and Kenya who label training data for a fraction of a living wage, the Phoenix data centers covering more than 450 football fields of land, and the power demands that may soon rival those of New York City.

What the S-1 will have to say about any of this is an open question. OpenAI has described its core mission in public statements as advancing artificial general intelligence that benefits humanity. The trial record now includes contemporaneous emails showing that same leadership could not define what that mission meant at founding. Whether the S-1 restates that language unchanged, revises it, or sidesteps it entirely is one of the more consequential things to watch when the document lands.

For institutional investors and underwriters, the question is not academic. Goldman Sachs and Morgan Stanley are underwriting an IPO for a company whose defining mission was, by its own founders' contemporaneous emails, undefined at founding. Any S-1 that restates that mission language unchanged inherits the definitional vacuum now on the public record. How the banks and their institutional clients account for that risk in their underwriting analysis is a question that has no clean answer — and one that will surface in the prospectus or in the roadshow meetings that follow.

A note on the sourcing: the Times reporting on Hao's vindication draws on her book Empire of AI, published in May 2025. The trial exhibits — the emails, memos, and text messages entered into the court record — are the new public-record anchor here. Those documents are not 13 months old. They are exhibit entries from a May 18, 2026 verdict.