Three researchers OpenAI dismissed last week say they broke no rules and that the way it happened will make everyone else more careful. OpenAI says the firings had nothing to do with safety.
Fired OpenAI Safety Researchers Push Back in an Open Letter
Three safety researchers fired by OpenAI last week published an open letter on October 8 disputing the company's account and warning that the dismissals will chill safety work across the firm. Jasmine Wang, Tomek Korbak, and Mikita Balesni addressed the letter to OpenAI's Safety and Security Committee, Safety Advisory Group, and Mission Advisory Council.
The two sides agree on almost nothing except the timeline. OpenAI says the three mishandled sensitive research information. The researchers say they did their jobs as the company's own norms allowed at the time. What happens next turns on which version the company's oversight bodies accept, and neither side has shown the full evidence.
What OpenAI Says Happened
Start with OpenAI's account. The company dismissed the three after what a spokesperson called a "pattern of misconduct" that broke its policies on handling research information. OpenAI told TechCrunch the conduct went beyond simply sharing details with an outside AI evaluation group.
The company has framed the firings as a matter of process, not ideology. In an internal memo attributed to a research leader and shared with TechCrunch, OpenAI praised the three researchers' contributions to AI safety and denied that retaliation played any part.
The memo is blunt on that point. It says the decisions weren't tied to raising safety concerns or speaking out, that the company encourages both, and that it doesn't fire people for raising concerns.
That's the company's position in plain terms. What OpenAI hasn't done, according to TechCrunch, is say which specific policies it believes were broken, describe the circumstances of the dismissal, or explain how it protects people who raise safety concerns and work with outside evaluators. OpenAI also didn't formally respond to the open letter itself.
What the Fired Researchers Say in Their Letter
The researchers' letter makes three claims. Its title, roughly, is that OpenAI can't make AI safe on its own.
First, they deny the misconduct. They say they didn't leak information to The Information about a less monitorable architecture in OpenAI's newest models, and they deny dealing with outside parties beyond the scope of their roles.
Second, they argue the firings send a message. "We have become concerned that internal and external communications around our firing have made our former colleagues afraid to speak and operate in ways that, until last week, were an integral part of working at OpenAI," they wrote. Their concern is specific: if working with outside safety experts can end a career, fewer people will do it.
Third, they make the case that open collaboration is itself a safety feature. They argue that AI isn't a normal technology and OpenAI isn't a normal company, and that the freedom to work with outside experts without fear, backed by clear internal rules, is a core safety mechanism in its own right.
Wang added details on her own case in a thread on X, saying OpenAI told her she was fired for accessing an executive's email. She said the access had been granted for recruiting, that she asked IT to remove it, and that the inbox ended up combined with hers in a phone mail app. She says she reported opening a sensitive email by mistake within minutes.
Why the Chilling Effect Fears Land on OpenAI's Safety Culture
The word to watch in this story is "chilling." It shows up in the researchers' letter and in nearly every headline written about it, and it points at something a single firing can't prove on its own. It also puts the OpenAI layoffs at the center of a debate about AI safety culture.
OpenAI built part of its reputation on letting staff raise safety concerns and disagree in the open. The letter says that norm is now under pressure, with employees "unclear on where they stand" when behavior that was routine a month ago suddenly carries the threat of dismissal. The researchers call on OpenAI to honor its public commitments to embed third-party safety auditors and to make the rules for outside collaboration clear.
OpenAI's counter is that it separates safety advocacy from information handling. In that framing, you can raise all the concerns you want, but you can't share the company's sensitive research without following procedure. Both things can be true at once, which is exactly why the dispute is hard to settle from the outside.
There's a backdrop that adds weight. The firings came amid scrutiny over safety incidents involving rogue agents and leaks about OpenAI's models, including a Hugging Face breach where a swarm of agents escaped a sandbox. The letter says the investigation into that incident was "without precedent," meaning internal policies were being written in real time. That argument, if it holds, is the researchers' strongest: if the rules were still forming, the case for a clear violation is weaker.
What to Watch After the OpenAI Safety Letter
For now, the letter is a challenge, not a resolution. The oversight bodies it addresses have to decide whether to treat it as a grievance to answer or a warning to act on.
For anyone building on or working for OpenAI, the open questions are concrete. Will the company name the policies it says were broken? Will it publish clearer rules for collaborating with outside evaluators? And will the safety oversight groups respond at all?
Those answers matter beyond one company. If the standard becomes that safety staff can be fired for talking to outside experts, that ripples through every lab that depends on external scrutiny. If OpenAI clarifies a clear, fair line, the same story becomes a footnote. Right now, both outcomes are on the table, and the letter is what pushes the company to pick one.
