Mira Murati, the former CTO of OpenAI, has testified that CEO Sam Altman lied to her about safety standards for a new AI model. In a video deposition during the Musk v. Altman trial, Murati revealed that Altman stated falsely that legal department approval was not required before deployment.

During her tenure at OpenAI, Murati described working under pressure and managing a complex organization. Her testimony highlighted discrepancies between Altman’s statements to her and those of Jason Kwon, the company's general counsel. To ensure safety, Murati insisted on going through the deployment safety board despite Altman’s assurances.

This is not the first time Altman has faced accusations of dishonesty; his co-founder Ilya Sutskever and former OpenAI board member Helen Toner have both described him as manipulative and disingenuous. The board that fired Altman cited a lack of consistent communication with them, impacting their ability to effectively govern the company.

Murati’s criticism extended beyond this case. She expressed concern about the potential for catastrophic failure at OpenAI under Altman's leadership and left the company in 2024 to found her own AI firm, Thinking Machines Lab. Her testimony may have significant implications for how we perceive and manage AI development.