OpenAI dismisses three AI safety researchers over alleged leak of sensitive information

OpenAI has dismissed three artificial intelligence safety researchers accused of leaking sensitive information, with the company citing several failures in how the information was handled, including the sharing of material with an outside AI evaluation group.

The company said the dismissed researchers had been involved in disclosing information to an external organization engaged in evaluating artificial intelligence systems, outside of approved internal processes.

OpenAI has not released the specific nature of the information shared or provided extensive public detail about the circumstances surrounding the dismissals, limiting its statement to describing the lapses that led to the decision.

The move comes amid continued attention to how major AI developers manage internal safety and security protocols, particularly for staff working directly on evaluating potential risks posed by advanced AI systems.

The dismissals add to broader industry discussion about the balance AI safety teams must strike between internal confidentiality requirements and engagement with outside researchers and evaluators.