The woman who ran OpenAI for three days after Sam Altman was fired just told a federal court exactly why the board did it.

Mira Murati, OpenAI's former CTO and briefly its interim CEO, delivered video testimony on Wednesday in the ongoing Musk v. Altman trial in Oakland. Her account was devastating. She told the court that Altman lied to her about whether a new AI model needed to go through OpenAI's deployment safety board. When she checked with OpenAI's then-general counsel Jason Kwon, their stories didn't match. Murati's words: "I confirmed that what Jason was saying and what Sam was saying were not the same thing."

Let that land for a second. The CEO of the company building what might be the most powerful technology in human history was allegedly lying to his own CTO about whether an AI model needed a safety review.

A Pattern, Not an Incident

Murati's testimony didn't stop at one lie. She agreed under questioning that Altman pitted executives against one another and undermined her ability to do her job as CTO. "I had an incredibly hard job to do in an organization that was very complex," she said. "I was asking Sam to lead, and lead with clarity, and not undermine my ability to do my job."

This is not a disgruntled employee venting on a podcast. This is sworn testimony from someone who had a front-row seat to OpenAI's internal operations for years, who was trusted enough to take over as CEO during the company's worst crisis, and who has since founded her own AI company, Thinking Machines Lab. She has nothing to gain from lying.

And Murati isn't alone. Ilya Sutskever, OpenAI's former chief scientist and co-founder, wrote in a 52-page memo to the board that Altman "exhibits a consistent pattern of lying, undermining his execs, and pitting his execs against one another." Former board member Helen Toner said executives had shared evidence of Altman "lying and being manipulative in different situations." The board itself, when it fired Altman in November 2023, said he "was not consistently candid in his communications with the board."

Three separate people. Three separate allegations. The same pattern.

The Safety Question That Should Keep Everyone Up at Night

Here's what makes this more than Silicon Valley drama. The specific lie Murati identified was about safety protocols. Altman allegedly told her that the legal team had cleared a model from needing safety board review. That wasn't true. The implications are staggering. If the CEO of the world's leading AI company is willing to bypass his own safety review processes, who exactly is watching the gate?

Murati, to her credit, overruled him. She made sure the model went through the board anyway. But the question isn't whether the system worked that time. The question is how many times nobody caught it.

What Happens Next

The trial continues. Altman is expected to testify next week, and you can bet every word of Murati's testimony will be put to him directly. OpenAI's defense has argued that Musk is a jilted co-founder seeking control he was never promised. That might be true. But Murati's testimony isn't about Musk's grievances. It's about whether OpenAI's CEO can be trusted with the most consequential technology of the 21st century.

So far, three of his closest former colleagues have answered that question under oath. And the answer is the same every time.

Sources: The Verge, Reuters, Forbes, NBC News, US News, NYT. Murati testified via video deposition on May 6, 2026 in Musk v. Altman, U.S. District Court, Northern District of California, Oakland.