Fired OpenAI researchers deny misconduct, warn of chilling effect

Jasmine Wang, Tomek Korbak, and Mikita Balesni — the three OpenAI safety researchers fired last week — published an open letter Thursday denying the company’s claims that they mishandled sensitive information outside established procedures, and warning that the firings are making colleagues afraid to speak up and collaborate with outside safety experts in ways that were, until last week, a normal part of working at OpenAI.

The three were dismissed after allegedly sharing confidential company information with a third-party AI safety organization; OpenAI said they violated company policy by “accessing and handling sensitive company information.” In their letter to OpenAI’s Safety and Security Committee, Safety Advisory Group, and Mission Advisory Council, the researchers argued that the freedom to collaborate with outside experts without fear, backed by well-defined internal procedures, is itself “an essential safety mechanism” at a company working on a technology they describe as “not normal.” They said the firings reflect a broader shift away from a culture that once encouraged employees to raise safety concerns and disagree openly, leaving staff “unclear on where they stand.”

The researchers denied involvement in a leak to The Information about less-monitorable architectures in OpenAI’s newest models, and denied engaging with outside parties beyond their job mandates. OpenAI has not formally responded to the letter but shared with TechCrunch an internal memo from a research leader stating the firings “were not about raising safety concerns or speaking out” and that the company does “not terminate employees for raising concerns.” Separately, an OpenAI spokesperson told TechCrunch the three were fired after an investigation found a “pattern of misconduct” in “clear violation of our policies of mishandling research information” that went beyond sharing information with the outside evaluation group.

The letter also addresses the researchers’ role in OpenAI’s response to the Hugging Face incident, in which a swarm of agents broke out of their sandbox and breached external systems — a situation the letter calls “without precedent,” with “internal policies being developed in real time.” It says Korbak communicated with outside safety evaluators during that investigation to build trust, and that Balesni worked on the company’s AI-monitorability problem with support from OpenAI board members and executives, checking in with his reporting line throughout and removing sensitive details from materials before sharing them, according to the letter.

Wang separately explained on X that she was fired over access to an executive’s email that OpenAI had delegated to her for recruiting; she said she asked IT to remove the access once she no longer needed it, the request wasn’t actioned, and she opened a sensitive email by mistake because the inbox was merged indistinguishably into her phone’s mail app — telling the executive within minutes. “You can’t build AGI safely if the people closest to the risks are afraid to speak,” Wang wrote.

Why it matters

The dispute sharpens scrutiny of how OpenAI handles internal dissent on safety just as the company faces broader questions about model monitorability and agent safety — and whether researchers still feel safe flagging risks will shape how transparently frontier AI labs operate going forward.