Three researchers who worked on safety and alignment at OpenAI were dismissed after the company said they mishandled sensitive information. The BBC reported that a spokesperson told the outlet on 2 October an internal probe confirmed the breach of established procedures. The former staff allege the firings were retaliation for raising safety concerns, while OpenAI maintains the decision was based on a trust violation.
Mikita Balesni posted on X that the dismissals resulted from prioritising safety over short-term corporate interests. The message accompanied an open letter addressed to OpenAI leadership, co-authored with Tomek Korbak and Jasmine Wang. All three described themselves as safety or alignment employees and asserted that their advocacy for robust safeguards prompted the termination.
OpenAI issued a note from its research leaders on Friday, reiterating that the terminations were unrelated to safety advocacy. The statement said an internal investigation uncovered a significant breach of trust beyond what the former employees described in their letter, and that the company stood by its decision not to retain them.
The episode arrives amid intensifying discussion about the societal risks posed by increasingly capable AI systems. Alignment work seeks to embed human ethical principles into machine behaviour, aiming to keep advanced models consistent with human values. Critics argue that corporate pressure to accelerate deployment can clash with such safety efforts.
In the joint letter, Balesni, Korbak and Wang warned that recent dismissals were creating a climate of fear among OpenAI staff. Korbak wrote that he had repeatedly raised concerns about losing the ability to monitor AI agents’ internal thoughts, a key tool for detecting misbehaviour. The authors said the environment had previously allowed open discussion of safety issues.
OpenAI indicated it is finalising contracts with third-party safety assessors and expects to announce details in the coming weeks. Wang added that the three researchers were not the first to leave under suspicious circumstances, and cautioned that without collective resistance, similar actions could recur in the industry.