- November 22, 2026
- Updated 8:36 am
Former OpenAI Researchers Dispute Dismissal Explanations
- 56 Views
- admin
- October 9, 2026
- Tech Companies Technology
Introduction
Three former safety researchers from OpenAI—Tomek Korbak, Jasmine Wang, and Mikita Balesni—have publicly challenged the company’s rationale for their dismissal. They expressed concern that the situation would dissuade other employees from voicing potential AI risks. Addressing the Safety and Security Committee, Safety Advisory Group, and Mission Advisory Council, the ex-employees emphasized the importance of transparency and collaboration.
Claims by Former Researchers
The researchers stated their dismay at how the dismissal communications have instilled fear among colleagues. They stressed that challenging decisions and collaborating with external experts are crucial in addressing AI safety concerns. Korbak, Wang, and Balesni argued, “AI is not a normal technology, and OpenAI is not a normal company.” They insist their dismissals were miscommunicated and that safety concerns should remain the priority.
Refuting the Reasons for Dismissals
The former researchers disputed OpenAI’s claims that policy violations led to their dismissal. Specifically, they denied responsibility for leaking sensitive information relating to AI architectures. Each researcher provided their account:
- Tomek Korbak: He was informed that his firing resulted from his communication style with external evaluators. Korbak argued he was merely performing his assigned duties and raised concerns about diminishing transparency in AI reasoning.
- Mikita Balesni: He stated that prioritizing safety over corporate interests was the real reason behind the dismissals. He maintained consistent feedback with management and removed sensitive material before external communication.
- Jasmine Wang: Her dismissal was attributed to allegedly accessing an executive email without permission. Wang explained that she was initially granted this access and sought to remove it upon completion of her responsibilities.
Recommendations from Former Employees
In their open letter titled “OpenAI cannot make AI safe on its own,” the researchers outlined several recommendations for OpenAI and other AI companies:
- Continue independent safety evaluations.
- Strengthen relationships with third-party organizations.
- Ensure transparency and allow safety staff to communicate with external experts freely.
Balesni further warned against OpenAI using their dismissals to cut ties with outside safety evaluators such as METR.
OpenAI’s Response
OpenAI defended its decision to release the researchers, citing a “significant breach of trust.” The company emphasized that the dismissals did not relate to raising safety concerns. OpenAI highlighted its culture of encouragement in critical safety discussions while protecting sensitive information.
Furthermore, OpenAI reaffirmed its dedication to independent safety evaluations and collaborating with external assessors. The company acknowledged the ex-employees’ contributions to AI safety and reassured ongoing commitment to transparency and open discourse.
Conclusion
The dispute underscores challenges in balancing AI development with safety protocols. Korbak, Wang, and Balesni stress the importance of open communication with external evaluators for effective oversight. Meanwhile, OpenAI insists on safeguarding sensitive information without stifling internal dialogues. The outcome of this conflict could influence future interactions between AI companies and safety researchers.