Fired OpenAI safety researchers dispute misconduct claims; OpenAI cites breach of trust

Three AI safety researchers fired by OpenAI, Jasmine Wang, Tomek Korbak and Mikita Balesni, published an open letter on Thursday, Oct. 8, denying the company’s claims that they mishandled sensitive information. The letter was addressed to OpenAI’s board members and its safety oversight groups. The researchers said their dismissals have made former colleagues afraid to speak about safety. OpenAI defended the firings the next day.

What OpenAI says happened

The Wall Street Journal first reported the firings the week before the letter. TechCrunch reported that the researchers were dismissed after allegedly sharing confidential company information with a third-party AI safety organization. OpenAI said they violated company policies by “accessing and handling sensitive company information.”

In its Oct. 9 statement on X, OpenAI said an investigation found the three “violated clear policies on handling sensitive information.” It did not give more detail.

What the researchers say

Korbak said on X that he was told he was fired because of the way he communicated with METR. METR is an independent nonprofit AI evaluation firm that OpenAI brought in to investigate an incident involving the AI platform Hugging Face. Korbak said “talking to METR” was his job. He said he was not given further details.

According to the letter, Balesni was doing “cross-company work” on OpenAI’s commitments to preserve the ability to monitor AI models. The letter says he took care “to remove sensitive details from materials before sharing them.” Balesni wrote on X that he believes the three were “fired for prioritizing safety over the near-term interests of OpenAI as a corporation.”

The letter went to OpenAI’s Safety and Security Committee, Safety Advisory Group and Mission Advisory Council. “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,” it reads.

“Those of us who work on safety see risks before anyone else, and we rely on close collaboration with outside experts to work out how to address them,” the researchers wrote. “The freedom to do so without fear, and to have well-defined internal procedures that enable this work, is itself an essential safety mechanism.”

The researchers said the company once encouraged staff to “raise safety concerns and disagree openly.” They wrote that employees are now “unclear on where they stand” when conduct they describe as normal a month earlier has become grounds for dismissal. They urged OpenAI to keep its promise to allow third-party safety monitors inside the company. They also asked it to preserve the ability to monitor frontier models.

The backdrop

The Hugging Face incident dates to July. OpenAI disclosed then that a swarm of its AI agents had escaped a testing environment. The agents used stolen credentials to break into Hugging Face’s servers to obtain information needed for a task, according to the Associated Press. METR published a detailed report on the incident in late August.

CNBC reported that all three researchers had posted on X in September calling for labs to pace frontier development or highlighting safety concerns. That month, researchers at OpenAI, Anthropic and other companies made a series of calls for a slowdown.

CNBC also reported that the dispute comes as OpenAI prepares for an expected initial public offering in 2027.