Mira Murati testifies about Altman

OpenAI’s governance mess is back in public view — and this time it’s coming out in sworn testimony, not leaked memos or postmortems. Mira Murati, OpenAI’s former CTO and the executive who briefly became interim CEO during Sam Altman’s 2023 ouster, told a federal court that Altman was not always truthful with her and that his management style created distrust at the top of the company. The immediate setting is Elon Musk’s lawsuit over OpenAI’s shift toward a for-profit structure, but the testimony matters beyond that case because it goes straight at the question that blew up the company in November 2023: was the board dealing with a normal hard-driving founder, or a CEO whose internal handling of safety and decision-making had become unmanageable? ### What did Murati actually say? In a recorded deposition played in federal court in Oakland on May 6, Murati said Altman was “not always” honest with her, undermined her as CTO, and told different people different things. She described her concern as Altman saying one thing to one person and “completely the opposite” to another, which she said created chaos among OpenAI’s top executives. ### What was the specific dispute? The sharpest allegation was about model safety review. Murati testified that Altman told her OpenAI’s legal department had concluded a new model did not need review by the company’s deployment safety board. When asked whether Altman was telling the truth when he said that, she answered: “No.” She said she checked with Jason Kwon — then OpenAI’s top lawyer, now chief strategy officer — and found that what Kwon was saying did not match what Altman had told her. Coverage tying this episode to GPT-4 Turbo says the dispute was one of the incidents that fed broader concerns about Altman inside the company. ### Why does that matter so much? Because this is not just a personality complaint. It goes to whether the company’s internal safety process could be bypassed, misrepresented, or selectively described depending on who was asking. OpenAI sells itself as an organization building unusually powerful systems while also taking unusual care around deployment. If a top technical leader says the CEO misled her about whether a model had to go through safety review, that cuts right into the credibility of that promise. ### How does this connect to the 2023 coup? Murati’s testimony fills in the human part of the November 2023 crisis. She said OpenAI was at “catastrophic risk of falling apart” when the board removed Altman, and that she worried the company could “completely blow up.” She was asked to step in as CEO during that scramble and stayed in close contact with Microsoft CEO Satya Nadella while the company tried to keep itself together. ### Did she want Altman gone? Not exactly — and that’s part of what makes the testimony more interesting. Murati described serious concerns about Altman’s behavior, but she also said she wanted him to remain CEO and pushed board members for a fuller explanation when they fired him. So this was not a clean “Murati versus Altman” story. It was more like: she thought his conduct was destabilizing, but also thought the board’s process risked detonating the whole company. ### Why is Musk’s lawsuit the venue for this? Musk sued OpenAI, Altman, and Greg Brockman in 2024, arguing that OpenAI abandoned its original nonprofit mission and became effectively a commercial engine tied to Microsoft. He is seeking $150 billion in damages, with Reuters saying the money would go to OpenAI’s charitable arm if he wins. Murati is not testifying to prove Musk’s whole case by herself, but her account helps Musk’s side paint OpenAI as a company whose internal controls and leadership discipline were weaker than advertised. ### So what changed this week? The big shift is that accusations that had circulated in profiles and insider accounts are now in court testimony. That does not mean every allegation is proven true. But it does mean the most damaging version of the governance critique — that Altman misled senior colleagues on safety and fostered distrust at the top — is now part of the formal record in one of the most consequential AI cases in the country. The bottom line is simple: Murati’s testimony does not just reopen old drama. It sharpens the core argument over OpenAI’s future — whether a company building frontier AI can ask for public trust while its own former top leaders describe the inside as chaotic, inconsistent, and hard to believe.