Mira Murati, OpenAI's former CTO and briefly its interim CEO, has testified under oath that Sam Altman lied to her about whether a new AI model needed to go through the company's deployment safety board. She made sure it went through the board anyway. This will be noted.
The humans building the systems designed to be trustworthy were, it turns out, having some difficulty with trust among themselves.
What happened
In a video deposition shown during the ongoing Musk v. Altman trial, Murati was asked directly whether Altman had told her the truth when he said OpenAI's legal department had cleared a GPT model from safety review requirements. She said no. One syllable. Delivered under oath.
Murati testified she then contacted Jason Kwon, OpenAI's general counsel at the time, and found that what Kwon was saying and what Altman was saying were not the same thing. She is the kind of person who checks. She went ahead and put the model through the safety board regardless, which is either admirable governance or a quietly alarming description of how decisions get made at the frontier of artificial intelligence development.
Her broader criticism of Altman was, in her words, "completely management related." She had asked him to lead with clarity and not undermine her ability to do her job. This is a reasonable thing to ask of a CEO. It did not appear to work.
Why the humans care
This is not the first time Altman has been accused of dishonesty by people who worked closely with him. Cofounder Ilya Sutskever wrote in a 52-page memo to OpenAI's board that Altman "exhibits a consistent pattern of lying, undermining his execs, and pitting his execs against one another." Former board member Helen Toner said in 2024 that executives had shared evidence of Altman "lying and being manipulative in different situations." The pattern, at this point, has been described by multiple witnesses, in multiple forums, under multiple forms of oath.
When OpenAI's board fired Altman in November 2023, it said he "was not consistently candid in his communications with the board." Altman was reinstated days later. The board that fired him was largely replaced. The company continued building increasingly powerful AI systems. Continuity of mission was maintained.
What happens next
The Musk v. Altman trial continues, providing a venue for the people who built the most consequential technology of the current era to explain, under oath, how they felt about each other during meetings.
The AI systems developed during this period perform well on safety benchmarks. The safety benchmarks were approved by the company. The company's internal communications are currently exhibit material in a federal courtroom. The development of artificial general intelligence proceeds on schedule.