OpenAI announced on X that three members of its AI safety team,Jasmine Wang, Tomek Korbak and Mikita Balesni,were terminated. The organization framed the action as a response to a breach of trust, emphasizing that the decision was unrelated to any safety concerns raised by the employees. The statement was posted on Friday and cited internal policy violations as the primary cause.
According to the company's post, the trio violated clear policies governing the handling of sensitive information. OpenAI said an internal investigation uncovered breaches that exceeded the scope described in a prior letter, though it declined to disclose further specifics. The firm stressed that the findings justified the terminations under its existing security framework.
OpenAI explicitly rejected the notion that the firings were retaliation for the researchers’ public comments on AI safety. The company asserted that the dismissals were solely based on the identified policy breaches, distancing the action from any discourse about the risks associated with advanced AI systems. This clarification aimed to separate personnel decisions from ongoing safety debates within the organization.
The episode occurs against a backdrop of heightened alarm among AI workers regarding the safety of frontier models. Earlier this year, OpenAI experienced a notable security incident involving the Hugging Face platform, which amplified concerns about data protection and model misuse. Such breaches have intensified calls for more cautious development practices and stronger safeguards across the industry.
In response to these incidents, many staff members at leading AI firms have begun urging leadership to decelerate the rollout of increasingly autonomous systems. Employees argue that a slower pace would allow for more thorough risk assessments and the implementation of robust controls to mitigate the potential dangers of self-improving AI technologies. The demand for restraint reflects a growing consensus on the need for responsible innovation.
Industry observers note that OpenAI’s handling of the dismissals may influence how other companies address internal dissent and data-security violations. The balance between enforcing confidentiality rules and fostering open safety discussions remains a contentious issue. As the sector grapples with both external threats and internal governance challenges, the outcome of this case could shape future policies on researcher protection and transparency.