Three former OpenAI researchers say they have been fired due to their efforts to ensure the safety of artificial intelligence. Safety researchers Mikita Balesni, Tomek Korbak, and Jasmine Wang leveled the accusations following their dismissals, warning of a chilling effect on other staff concerned about the risks of frontier technology. OpenAI maintained that they were let down due to misconduct involving the mishandling of company information.
Balesni stated on social media that he believes he was terminated for prioritizing safety over the near-term corporate interests of OpenAI. Korbak expressed a similar sentiment, stating he believes he was fired for raising concerns that the company was losing its ability to monitor what artificial intelligence agents think, which serves as a crucial tool for catching misbehavior. In an open letter published alongside their social media statements, the trio warned that their dismissals would make former colleagues afraid to speak openly and operate in ways that were previously integral to the company's culture.
The researchers denied violating company policy, stating that any engagement with external safety experts fell entirely within the mandate of their roles. They argued that employees must not be forced to work in an environment where fear and ambiguous rules hinder critical safety work and weaken third-party accountability. They added that abrupt terminations of this nature chill the open culture that OpenAI has historically championed.
In response, OpenAI stated that it uncovered a significant breach of trust by the former researchers that went far beyond what was described in their letter, and emphasized that it stands fully behind the termination decisions. The San Francisco-based enterprise stated clearly that the actions were unrelated to raising safety concerns or speaking out, noting that spirited and critical safety debates occur daily within the organization.
OpenAI added that while it appreciates contributions to AI safety and encourages dissenting voices, the work requires an exceptionally high degree of mutual trust. The controversy arrives amid broader industry debates regarding how to ensure rapidly evolving technology does not expose humanity to catastrophic harm. In the absence of formal regulation, OpenAI has taken steps to mitigate risks, including scrapping the planned release of its latest-generation model after it failed to meet company standards for aligning with human intentions.