OpenAI fires three safety researchers after they raised concerns about model monitorability

OpenAI terminated researchers Jasmine Wang, Tomek Korbak, and Mikita Balesni, stating the decision was due to a 'significant breach of trust' and not related to their safety concerns. The researchers had urged OpenAI to pause development of models that reduce AI monitorability in a letter to the board and safety committees. OpenAI said it agrees with the letter's goal of preserving monitorability and continues to invest in this area.

Why it matters. The incident highlights internal tensions at a leading AI developer over safety practices and transparency, potentially affecting trust in AI governance and researcher freedom to raise concerns.