Mira Murati testifies in OpenAI trial

OpenAI’s courtroom fight with Elon Musk just got much more concrete. Instead of abstract arguments about nonprofit charters and AGI ideals, the jury is now hearing former insiders describe how decisions were made inside the company. Mira Murati’s testimony matters because she was not a bystander — she was OpenAI’s chief technology officer, and she sat close to both the product pipeline and the leadership drama. What changed this week is that complaints people had mostly read in profiles and leaks are now being aired under oath. (usnews.com) ### What did Murati actually say? Murati testified by video on May 6 in federal court in Oakland, where Musk is trying to prove OpenAI abandoned the founding mission he says he backed at the start. Her account was blunt: she said Altman created distrust(usnews.com)safety process, not personality. (usnews.com) ### What was the safety dispute? Murati said Altman told her OpenAI’s legal team had concluded a new model did not need review by the company’s deployment safety board. She testified that this was not true. After checking with then-legal chief Jason Kwo(usnews.com)It is “the CEO may have tried to route around a formal safety checkpoint.” (gizmodo.com) ### Why is that detail such a big deal? Because Musk’s whole case depends on showing that OpenAI’s structure changed its behavior. He argues the company moved away from its original public-interest nonprofit mission and toward a more commercial, closed model tied to Microsoft and product speed. (gizmodo.com)sophical dispute into a governance dispute — who had authority, what checks existed, and whether those checks were bypassed. (techcrunch.com) ### Is this just about the 2023 Altman firing? Not really — but that crisis is hanging over everything. Murati also said concerns about Altman did not disappear after he returned as CEO in November 2023. That matters because OpenAI has often framed the board revolt as a one-off breakdown by (techcrunch.com)ave lasted longer than that story implies. (bizjournals.com) ### Where does Shivon Zilis fit in? Zilis, a former OpenAI board member and longtime Musk associate, also testified during the trial’s second week. Her appearance pulled attention toward Musk’s ties to OpenAI during its earlier years and to questions about who knew what inside the company(bizjournals.com)text. Murati adds operational detail. (abc7news.com) ### Does this prove Musk wins? No. Testimony is evidence, not the verdict. OpenAI can still argue that internal disagreement is normal at a fast-moving lab, that safety review happened in the end, and that Musk’s lawsuit overreaches. The catch is that juri(abc7news.com)corporate form changed.” (techcrunch.com) ### Why should anyone outside the trial care? Because this is becoming a real test of AI governance, not a Silicon Valley soap opera. Frontier labs keep saying they take safety seriously. This trial is forcing one of the biggest labs to show what that meant in practice — who signed off, who (techcrunch.com)ystems only matter if they can survive pressure from the top. (techcrunch.com) ### Bottom line? Murati’s testimony does not settle the case. But it gives Musk’s lawsuit its clearest factual hook yet. The argument is no longer just that OpenAI changed. It is that, inside the company, speed and executive power may have started outrunning the safeguards that were supposed to make the whole project trustworthy. (gizmodo.com)