Mira Murati says Sam Altman misled her about OpenAI safety review

In short: Former OpenAI CTO Mira Murati told a court that Sam Altman lied to her about whether a new AI model needed a safety review.

What happened

Mira Murati, who used to be OpenAI’s chief technology officer, testified under oath in the Musk v. Altman trial. Her testimony was shown in court as a video deposition, which is recorded sworn testimony.

Murati said Altman told her that OpenAI’s legal team had decided a new AI model did not need to go through the company’s deployment safety board. A deployment safety board is an internal review group that checks whether a model is safe enough to release, like a final inspection before a product ships.

When asked directly if Altman was telling the truth, Murati answered “No.” She said she later checked with Jason Kwon, OpenAI’s top lawyer at the time, and found that Kwon’s account did not match Altman’s. Murati said she made sure the model went through the safety board anyway.

Murati also criticized Altman’s management. She said the issue was not technical, but about leadership and clarity. She told the court that Altman made her work more difficult and, in her view, undermined her ability to do her job.

Why it matters

Many people use AI tools in everyday products, from chatbots to writing and image apps. When an AI company skips or weakens internal checks, it raises questions about who is responsible if something goes wrong. Murati’s testimony adds to the pressure on OpenAI’s leadership as the court considers arguments about how the company should be run and what rules should govern its AI releases.

Source: The Verge AI