The three former OpenAI researchers who were suddenly dismissed last week — Mykita Balexnier, Tomek Kolbakh and Jasmin Wang — formally published an open letter today. In the letter, they stated plainly that they worry this abrupt move by the company will leave other current employees too afraid to speak up, dealing a serious negative blow to OpenAI's overall corporate culture.
The open letter, titled 'OpenAI Cannot Ensure AI Safety on Its Own,' was published on social media by Balexnier, who previously worked in AI alignment research, and was addressed to OpenAI's various safety committees. The two other signatories, Kolbakh and Jasmin Wang, likewise previously focused on AI safety research and were dismissed at the same time last week.
As for the specific reason for the dismissals, Balexnier revealed in his post that the company's stated justification was a breakdown of trust due to his 'excessive communication with third-party safety organizations.' He clarified that he had never leaked any of the company's intellectual property externally, and that his daily work had always been closely coordinated with his direct supervisor, the research division's management, and the board. Another signatory, Kolbakh, said the company verbally told him he was dismissed over 'the way he communicated with METR,' but he emphasized that communicating with METR was part of his job duties.
In response to this turmoil, the three expressed deep concern in the open letter, arguing that this sudden and hasty manner of dismissal has subjected OpenAI's once-proud culture of openness to a chilling effect, and may lead other employees to no longer dare to freely express opinions or carry out their work as before. To this end, the three made three specific recommendations to OpenAI: first, allowing third-party safety auditors to be stationed at the company; second, ensuring that frontier AI models remain under effective continuous monitoring; and third, supporting safety researchers in maintaining open and transparent communication with other organizations in the industry.
In the face of outside skepticism, a head of OpenAI's research division expressed strong support for these three recommendations in an internal memo subsequently sent to employees. The executive said the company greatly appreciates the three researchers' contributions to AI safety work and their courage to speak out, and clarified that the dismissal decisions had nothing to do with their raising safety concerns or speaking publicly, emphasizing that OpenAI has always encouraged — and will continue to encourage — employees to express their opinions, and will never dismiss employees for raising concerns.