Former OpenAI CTO Mira Murati Testifies CEO Sam Altman Deceived Her on AI Safety Protocols
Murati states Altman falsely claimed a new model did not require board review, a discrepancy confirmed by General Counsel Jason Kwon.
Mira Murati, the former chief technology officer of OpenAI, has given sworn testimony in the Musk v. Altman trial alleging that CEO Sam Altman provided her with false information regarding safety standards for a new artificial intelligence model. During the deposition, Murati stated that Altman told her the company's legal department had determined the model did not require review by the deployment safety board.
Upon verifying this claim, Murati confirmed a significant misalignment between Altman's assertions and those of General Counsel Jason Kwon. She testified that after speaking with Altman, she consulted Kwon to ensure the matter was handled correctly. Finding that the two executives were not in agreement, Murati proceeded to ensure the model underwent the necessary review by the deployment safety board to mitigate risk.
Murati characterised the situation as being completely management related, noting that Altman's actions undermined her ability to perform her duties effectively. She described the environment as complex and stated she was asking Altman to lead with clarity rather than creating obstacles for the executive team. Her criticism focused on the difficulty of leading an organisation while facing inconsistent communications from the top.
This testimony adds to a broader narrative of internal discord at OpenAI, with previous accusations from co-founder Ilya Sutskever describing a consistent pattern of Altman lying and pitting executives against one another. Former board member Helen Toner has also noted in public forums that executives shared evidence with the board regarding Altman being manipulative in various situations.
The board's decision to dismiss Altman in November 2023 was driven by a loss of confidence in his candour and inconsistent communications. Murati acknowledged these descriptions, agreeing with the characterisation of Altman undermining her role. The board had previously stated that Altman was not consistently candid, hindering their ability to exercise their responsibilities and leading to a lack of trust in his leadership.
Murati was briefly appointed as interim CEO following Altman's dismissal but later departed OpenAI in 2024 to found Thinking Machines Lab. Her current testimony in the trial underscores the tensions that existed during her tenure and the specific governance failures she identified regarding the safety protocols for the company's GPT models.
