OpenAI has quietly cut three safety researchers loose, and the reason sounds more like a data-handling violation than a culture clash.
The Wall Street Journal reports that OpenAI dismissed three employees from its safety organization after an internal investigation found they mishandled sensitive company information. The report doesn't specify what information was involved, how it was mishandled, or whether it left the company. It also doesn't name the researchers or detail their specific roles in OpenAI's safety work. Beyond that, there's little public detail to go on.
Any exit from OpenAI's safety ranks draws outside scrutiny, given the company's history of tension between shipping fast and policing its own models. Framing this as an information-handling problem rather than a policy dispute is notable: it points to an internal-controls failure, not a disagreement over AI risk, though with safety staff specifically, the line between the two is never as clean as a press release makes it sound.
Until OpenAI says more, these firings will get read through the lens of every prior safety departure at the company, fair or not.