OpenAI responds to fired employees' letter, defends termination

Tomek: Tomek Korbak is a safety researcher formerly at OpenAI with a background in reinforcement learning for model alignment. He was terminated by the company last week and co-signed the open letter addressing the handling of their departures and advocating for continued external collaborations on AI safety. Mikita: Mikita Balesni is a safety and alignment researcher who worked at OpenAI until his recent termination. Along with two colleagues, he signed the open letter to OpenAI leadership pushing back on misconduct allegations and highlighting concerns about a potential chilling effect on internal safety discussions. OpenAI: OpenAI is an AI research and deployment company developing advanced artificial intelligence systems. In this news, the company publicly responded to an open letter from three former safety researchers it terminated last week, clarifying that the decision stemmed from violations of policies on handling sensitive information rather than retaliation for raising safety concerns. Jasmine: Jasmine Wang is a safety researcher previously employed at OpenAI, where she contributed to the safety cases program. She was one of three employees recently parted ways with by the company and co-authored the open letter disputing the stated reasons for their terminations while emphasizing the importance of collaboration with external safety groups. Research Culture: OpenAI maintains that safety and research debates occur daily within the company, are actively encouraged, and that employees are not terminated for raising concerns in good faith. Safety Partnerships: OpenAI is actively finalizing contracts with third-party safety assessors and plans to announce details in the coming weeks as part of embedding external oversight in its work. Model Monitorability: OpenAI agrees that preserving the monitorability of frontier models requires an industry-wide commitment and continues to invest in related research programs and publications.