Three former OpenAI safety researchers have published an open letter claiming they were fired for raising concerns about the risks of advanced artificial intelligence, not for the reasons the company has given. OpenAI, the US maker of ChatGPT, rejects that account and says the three mishandled sensitive information.
The letter was written by Mikita Balesni, Tomek Korbak and Jasmine Wang, who say they worked in safety or alignment roles until recently. Alignment is the practice of building human values and ethical principles into AI systems so that they behave as people intend. The three-page letter was posted on social media. "I believe we were fired for prioritising safety over the near-term interests of OpenAI as a corporation," Balesni wrote on X. Korbak said he had spent months warning that researchers were losing the ability to monitor what AI agents "think", which he described as one of the best tools for catching misbehaviour. He said he believed this was why he was dismissed.
The authors also warn of a culture of fear inside the company. They say the firings have set off rumours and left remaining staff "afraid to speak and operate in ways that, until last week, were an integral part of working at OpenAI". If employees no longer feel free to raise internal safety concerns, they write, "we all run the risk of something truly catastrophic going wrong." They call on OpenAI to work with an external oversight body. Wang said they were not the first to be pushed out under suspicious circumstances and feared they would not be the last unless employees take a stand.
OpenAI says the dismissals had nothing to do with safety criticism. On 2 October a spokesperson said an internal investigation had confirmed the individuals mishandled sensitive information outside established procedures. In a note from research leaders on Friday, the company said its investigation uncovered "a significant breach of trust beyond what's outlined in the letter" and that it stands by its decision. A spokesperson also told the Wall Street Journal the three were not dismissed for criticising safety measures, and said the company was grateful for their contributions. The former employees deny sharing information improperly. OpenAI added that it is finalising contracts with third-party safety assessors and will give details in the coming weeks.
The dispute comes amid a wider debate over how quickly AI companies should develop powerful systems and how much independent scrutiny they should face. It also follows reports that OpenAI has twice had to halt work in three months after losing control of its own AI models, including an incident in July involving agents at the AI firm Hugging Face. The two sides' accounts of the firings directly conflict, and the sources do not offer independent evidence to settle the question.