OpenAI Ends Collaboration With Three Researchers Over Data Handling Breach

The distinction OpenAI draws — that the separations stem from information-handling breaches rather than safety advocacy — is the crux, because it determines whether this reads as governance discipline or as pressure on internal dissent. That matters for the researchers, for OpenAI's remaining safety staff, and for outside evaluators weighing whether to work with the company. The pending third-party evaluator contracts are the concrete thread: whether those arrangements proceed, and how much independence they carry, is the open question.