OpenAI Fires Three Safety Researchers Over Info Shared With Outside Groups
The company says the researchers mishandled sensitive information by sharing it with outside AI-safety organizations, a move that has revived questions about how OpenAI treats staff who raise safety concerns.
OpenAI has fired three researchers on its safety team after an internal investigation concluded they shared sensitive company information with outside AI-safety organizations, according to reporting by The Wall Street Journal that other outlets confirmed on Thursday.
The company has not said what the information was. It told reporters the employees were let go for violating policies on accessing and handling sensitive material. "OpenAI takes the safeguarding of its research and proprietary information extremely seriously," it said, adding that sharing such material without authorization breaks company policy regardless of who receives it.
The information went to organizations focused on AI safety research and advocacy, not to news outlets or competitors, according to the reports. The researchers have not been publicly identified in several of the accounts, and OpenAI has declined to name them.
The firings land at a sensitive moment for the industry, as concern about AI agents behaving in unexpected ways has grown in recent weeks. After earlier security incidents, OpenAI let staff from outside evaluators such as METR and Redwood Research look at how its models work.
OpenAI has not said what was shared, so it is unclear whether the material concerned model capabilities, safety testing or research findings. The company has also not said whether the outside organizations used the information or published any of it.
The episode echoes earlier fights over safety and speech inside the company. In 2024, OpenAI fired researcher Leopold Aschenbrenner after he shared safety documents with outside researchers. Jan Leike, who co-led the company's superalignment team, resigned that year saying safety culture and processes had "taken a backseat to shiny products." The team was later dissolved.
Some observers say the dismissals send the wrong message. They note that chief executive Sam Altman has publicly said the industry should pace the development of powerful AI, while his company is disciplining researchers who sought outside input on safety. OpenAI's position is that its rules protect trade secrets and that internal review, not outside disclosure, is the proper route for concerns. The company's statement did not address whether the dismissed researchers had raised their concerns internally first, and it did not say whether other employees are under review.
It is unclear whether the fired researchers will challenge their dismissals or whether lawmakers will ask questions about the company's whistleblower policies. The Federal Trade Commission opened an investigation this week into OpenAI, Anthropic and other AI firms over consumer risks, which adds regulatory scrutiny to a company already under pressure over how it handles safety.
