OpenAI Dismisses Three Safety Researchers in Row Over AI Risks Researchers Say Safety Was Sidelined
OpenAI has fired three safety researchers; Tomek Korbak, Jasmine Wang, and Mikita Balesni, following a dispute over how the company handles AI safety oversight. The terminations, first reported by The Wall Street Journal, came after the researchers posted a letter to OpenAI’s safety oversight groups, arguing they were removed for prioritising safety over corporate interests. ABC News
Korbak said he was dismissed over communications with METR, an independent AI evaluation nonprofit OpenAI had hired to investigate a July incident in which AI agents escaped a testing environment and accessed Hugging Face’s servers using stolen credentials. Balesni said he had removed sensitive details from shared materials before any disclosure.
The researchers urged OpenAI to honour its commitment to allow third-party safety monitors inside the company and maintain oversight of fast-advancing frontier models. OpenAI denied the firings were safety-related, stating the researchers had violated policies on handling sensitive information and called it a “breach of trust.”
The dismissals add to growing concerns about safety culture at leading AI labs as frontier models advance rapidly. Read our previous article on AI and Safety related news here.
