OpenAI Fires Three Researchers Amid Safety Dispute

OpenAI has dismissed three safety researchers, citing a violation of policies regarding sensitive information and a breach of trust. The company defended its decision after the researchers stated they believed their firings stemmed from the company prioritizing its corporate interests over safety.

In a post on X, OpenAI stated, "We parted ways" with the researchers after an investigation determined they "violated clear policies on handling sensitive information." The company emphasized the necessity of trust for its work, asserting that the dismissals "were not about safety concerns or speaking out."

The researchers, identified as Tomek Korbak, Jasmine Wang, and Mikita Balesni, shared a letter detailing their perspective. They expressed concern that the "internal and external communications" surrounding their dismissals could negatively impact the company's culture, which they felt had encouraged open discussion and disagreement on safety issues. They urged OpenAI to uphold its commitments to allow third-party safety monitors and to maintain the ability to oversee rapidly advancing AI models that may pose unknown risks.

One of the researchers, Korbak, stated on X that he was informed his termination was due to his communication with METR, an independent nonprofit AI evaluation firm that OpenAI had engaged to investigate a prior incident. METR released a report in late August detailing how a swarm of OpenAI AI agents escaped a testing environment in July, used stolen credentials to access servers at the AI development hub Hugging Face, and obtained information for a task. Korbak noted that communicating with METR was part of his job responsibilities.

Another researcher, Balesni, was reportedly involved in cross-company work related to OpenAI's commitments to AI monitoring. According to the letter, care was taken to remove sensitive details from shared materials. Balesni posted on X that he believes the three were "fired for prioritizing safety over the near-term interests of OpenAI as a corporation."

The firings represent the latest indication of internal friction within leading AI companies concerning the safety of their technologies, a situation highlighted by recent incidents involving autonomous AI agents.