Former OpenAI Safety Researchers Say They Were Fired Over Safety

Three former OpenAI safety researchers said they were fired after putting model safety ahead of OpenAI’s short-term corporate interests, prompting a call for stronger independent oversight of frontier artificial intelligence systems.

Mikita Balesni, Tomek Korbak and Jasmine Wang sent a letter to OpenAI leadership disputing the circumstances of their dismissals. The researchers denied leaking confidential company information.

They urged OpenAI to follow through on its commitment to introduce third-party safety audits and to preserve the ability to monitor its most advanced models.

OpenAI previously said an internal investigation found that the three researchers handled sensitive information outside established procedures. The company’s statement contrasts with the researchers’ account that their focus on safety contributed to their dismissals.

The dispute places internal safety processes and external review commitments at the center of a wider disagreement between the former researchers and OpenAI leadership.