Former OpenAI CTO Mira Murati has testified under oath that CEO Sam Altman lied to her about safety standards for a new AI model. In a video deposition presented during the ongoing Musk v. Altman trial, Murati stated that Altman falsely claimed OpenAI’s legal department had determined that a new AI model didn’t need to undergo review by the company’s deployment safety board. When asked if Altman’s statement was truthful, Murati replied, “No.”
Murati stated Altman made her job more challenging, criticizing his management style. She mentioned the complexity of working at OpenAI and expressed a need for Altman to lead clearly.
A specific example involved discussions about OpenAI’s GPT model. Murati consulted Jason Kwon, OpenAI’s general counsel, and found discrepancies between Kwon’s and Altman’s statements. To ensure safety, she insisted the model go through the board.
Altman has faced previous lying accusations. Cofounder Ilya Sutskever accused Altman of a pattern of deceit in a memo to OpenAI’s board. Former board member Helen Toner confirmed providing evidence of Altman’s manipulative behavior. Murati agreed Altman pitted executives against each other.
OpenAI’s board, when dismissing Altman, cited a lack of trust in his leadership due to inconsistent candor. Altman, when questioned, suggested asking the board why trust was lost.
Murati, briefly appointed interim CEO after Altman’s firing, criticized the board’s decision, fearing OpenAI’s collapse. She departed OpenAI in 2024 and founded her company, Thinking Machines Lab.