OpenAI Says Safety Matters as Fired Researchers Accuse It of Closing Ranks

OpenAI Says Safety Matters as Fired Researchers Accuse It of Closing Ranks
OpenAI’s dispute with three dismissed safety researchers exposes a widening fault line: whether the company’s pursuit of powerful AI can coexist with independent scrutiny and dissent from inside its own ranks.

The researchers—Tomek Korbak, Jasmine Wang and Mikita Balesni—announced their terminations after what they described as a conflict over how safety work should be conducted. They suggested they were fired for collaborating with outside experts to develop safeguards, raising the broader concern that OpenAI is becoming less receptive to independent oversight and researchers who challenge corporate decisions.

Their account frames the dismissals as more than an internal personnel dispute. The central allegation is that OpenAI is narrowing the space for external checks just as the risks associated with increasingly capable AI systems are becoming harder to assess from inside one company. A separate report characterized the firings as a “breach of trust,” underscoring the seriousness of the researchers’ claims.

OpenAI offered a sharply different interpretation. The company defended the dismissals and rejected the implication that it opposes safety research or outside scrutiny. It said safety and research debates are “highly critical” to its success, presenting disagreement as a necessary part of its work rather than evidence of a retreat from oversight.

The two positions overlap on one point: AI safety remains central. They diverge over who gets to define that safety and how openly the company should be challenged. For the researchers, collaboration beyond OpenAI appears to have become risky; for the company, the firings do not invalidate its stated commitment to debate. That unresolved contradiction is the real story—and a test of whether OpenAI’s safety culture can withstand scrutiny from beyond its corporate boundaries.