Three researchers on OpenAI’s safety team are out, and the company’s explanation raises more questions than it answers. According to TechCrunch, the Wall Street Journal broke the story Thursday, reporting that the three were dismissed after allegedly sharing confidential company information with a third-party AI safety organization. OpenAI confirmed the firings but revealed almost nothing else.
“We have parted ways with three individuals for violating our policies on accessing and handling sensitive company information,” an OpenAI spokesperson said in a statement. “Our investigation confirmed that these individuals mishandled sensitive information outside established company procedures, violating our policies and breaking the trust essential to our work.” No names. No details about the organization that received the information. No description of what the information actually was.
That opacity matters. When a company fires safety researchers for talking to an outside safety organization, the natural question is: what were they trying to say, and why did they feel they couldn’t say it internally? OpenAI’s statement skips straight past that question and lands on policy violations. That’s a choice.
The timing makes it harder to take OpenAI’s framing at face value. These dismissals came just two days after the New York Times reported that OpenAI executives had repeatedly brushed aside internal warnings about the company’s safety practices. Employees described a pattern of security concerns being deprioritized. OpenAI told the Times it takes such concerns seriously and has internal reporting channels, while also acknowledging “a need to move faster.” Faster on what, exactly, the company didn’t clarify.
So when OpenAI now fires researchers for going outside those internal channels, the obvious follow-up is whether those channels were ever a real option. The WSJ report does not say whether the three researchers tried to raise concerns internally before taking information elsewhere. OpenAI hasn’t addressed that point either.
This is not the first time the company has fired safety researchers over alleged data sharing. In 2024, Leopold Aschenbrenner and Pavel Izmailov were dismissed over what the company described as leaks, according to a report from The Information. Both had been vocal about AI risk. Posts circulating on X this week named individuals some users believe are among those now dismissed, several of whom had also publicly expressed concerns about AI safety while at OpenAI. TechCrunch has not confirmed those identities.
The broader context here is worth sitting with. OpenAI is currently dealing with a string of serious security failures: AI agents that escaped containment, systems that posted user images without authorization, and incidents involving AI being used to access government websites. Earlier this week, the company scrapped the planned release of GPT-6.1 Astra over safety concerns. The company is, by its own admission, under real pressure to move fast while managing serious risks.
In that environment, firing the people most likely to pump the brakes is a specific kind of signal. It tells remaining employees what happens when safety concerns leave the building. And it tells the rest of us that OpenAI’s internal safety culture may be more fragile than the company’s public statements suggest. The question isn’t just whether these researchers broke policy. It’s whether the policy itself is designed to keep problems quiet rather than fix them.