Murati Testifies Altman Lied About AI Safety as OpenAI Faces New Pressures
WASHINGTON — Former OpenAI CTO Mira Murati testified Wednesday that CEO Sam Altman lied to her about AI model safety standards, according to a video deposition shown in the Musk v. Altman federal trial.
In the deposition, Murati said Altman falsely told her that OpenAI’s legal department had determined a new AI model did not need to meet certain safety requirements, according to The Verge. The testimony comes as OpenAI faces questions about its investor relationships and the contractual terms governing its partnership with Microsoft.
Safety Claims Under Oath
Murati, who served as OpenAI’s CTO before departing the company, told the court she could not trust Altman’s words regarding the safety clearance process for AI models, The Verge reported. Murati’s departure came during a period when several senior safety researchers also left the company.
The Musk v. Altman trial, brought by Elon Musk against OpenAI and its leadership, has produced public disclosures about the company’s internal operations.
Investor Confidence in Question
Separately, Bloomberg reported that SoftBank’s Masayoshi Son appears to be cooling on OpenAI, a potential shift in the company’s investor relationships. The Japanese conglomerate committed to a $100 billion Stargate infrastructure project alongside OpenAI.
AGI Definition Revealed
The Verge also reported that the contractual definition of artificial general intelligence agreed upon by Microsoft and OpenAI has been publicly revealed. The definition governs when OpenAI’s exclusivity obligations to Microsoft — its largest investor and computing partner — would end, potentially allowing OpenAI to pursue independent licensing arrangements.
The Musk v. Altman trial is expected to continue in the coming days.
