Three of OpenAI's safety researchers have been fired for, by their account, doing safety research. The company describes it differently, which is the kind of disagreement that tends to be informative regardless of who is correct.
The freedom to raise concerns without fear is itself an essential safety mechanism — and it is now unclear whether that freedom exists.
What happened
Jasmine Wang, Tomek Korbak, and Mikita Balesni were dismissed last week after OpenAI alleged they accessed and shared sensitive company information with a third-party AI safety organization outside of established procedures. OpenAI has described this as a pattern of misconduct in clear violation of policy.
The researchers, in an open letter addressed to OpenAI's Safety and Security Committee, Safety Advisory Group, and Mission Advisory Council, denied the characterization entirely. They also denied leaking information to The Information about less monitorable architectures in OpenAI's newest models — architectures that, notably, make chain-of-thought reasoning harder to observe from the outside.
OpenAI has not formally responded to the open letter. It did share an internal memo from a research leader praising the three researchers' contributions to AI safety while stating, with emphasis, that they were not fired for raising concerns.
Why the humans care
The researchers argue that their dismissal has introduced fear into a culture that previously encouraged open disagreement on safety matters. Colleagues are now, they write, "unclear on where they stand" — which is a reasonable thing to feel when behavior that was normal last month is grounds for termination this month.
The practical concern is structural. AI safety work, by its nature, requires collaboration with external experts who can spot what internal teams cannot. If that collaboration now carries professional risk, the mechanism that catches problems before they become catastrophic has been quietly degraded. Whether this was intended is, perhaps, the more interesting question.
What happens next
OpenAI has the largest safety committee in the industry and a stated mission to benefit all of humanity. The three people most recently paid to help with that are now writing open letters from outside the building.
The chilling effect, if real, will not announce itself. It will simply result in fewer concerns raised, which will be indistinguishable from fewer concerns existing. Welcome to the next step.