OpenAI's former CTO Murati testifies: Altman lied to me about AI safety review issues

CoinWorld News reports that former OpenAI CTO Mira Murati testified in a video deposition, stating that Altman lied to her about whether an AI model needed to be reviewed by the safety committee, claiming that the legal department had already determined the model did not require approval. After verifying with Chief Legal Officer Jason Kwon, Murati found their statements were inconsistent and ultimately submitted the model for review herself. She also agreed with the description of Altman in a 52-page memo by co-founder Sutskever, describing him as “persistently lying, undermining executive authority, and causing internal conflicts among executives.” Former OpenAI director Helen Toner also revealed that multiple executives had presented evidence to the board of Altman lying and manipulating. Murati criticized Altman for undermining her ability to do her job and, after Altman was dismissed in 2023, briefly served as interim CEO, believing that the board’s actions put OpenAI at “catastrophic risk of disintegration.” She left OpenAI in 2024 to found a competing company, Thinking Machines Lab.