Former OpenAI researchers say safety concerns led to firings

Three former OpenAI researchers say the company fired them for raising safety concerns, a claim OpenAI rejects. The company says an internal investigation found they mishandled sensitive information and committed a significant breach of trust.

Mikita Balesni, Tomek Korbak and Jasmine Wang made their accusations in an open letter to OpenAI’s leadership and in posts on X. They said their dismissals could make other employees afraid to speak openly about risks posed by artificial intelligence.

“I believe we were fired for prioritising safety over the near-term interests of OpenAI as a corporation,” Balesni wrote.

OpenAI told the BBC on October 2 that its investigation “confirmed that these individuals mishandled sensitive information outside established company procedures”. In a subsequent note from research leaders, the company said the investigation had uncovered “a significant breach of trust beyond what’s outlined in the letter they published”.

“We want to be very clear that these decisions were not about raising safety concerns or speaking out,” the note said. OpenAI said it stood by the decision to end their employment.

The three former employees said they had worked in safety or alignment roles, which focus on keeping AI systems consistent with human values. Korbak said he had spent months warning that OpenAI was “losing the ability to monitor what AI agents think, one of our best tools for catching when they misbehave”. He said he believed that concern was why he was fired.

In their letter, the researchers said they had previously been able to disagree openly and draw on the expertise of independent safety organisations. They warned that their dismissals had made colleagues “afraid to speak and operate in ways that, until last week, were an integral part of working at OpenAI”.

“Terminations such as ours, executed and communicated so abruptly, are chilling the open culture OpenAI has prized in the past,” they wrote. They also said safe AI development depends on people closest to the risks being able to work closely with one another and with outside experts.

The three denied violating company policy, saying their contact with external safety experts was consistent with their jobs, according to Al Jazeera. Wang wrote on X that the group was not the first to be pushed out of OpenAI “under suspicious circumstances” and said she feared others could follow.

OpenAI said it encourages spirited, critical safety debates and remains willing to forgive good-faith mistakes. It said it appreciated the researchers’ contributions and their willingness to challenge ideas, while maintaining that trust was essential to its work.

The dispute comes amid wider debate over the risks of advanced AI and how companies should assess them. OpenAI said it was finalising contracts with third-party safety assessors and would announce details in the coming weeks, the BBC reported.