OpenAI dismisses three safety researchers accused of sharing confidential material

Three OpenAI safety researchers have been dismissed for allegedly mishandling confidential company information and sharing it with a third-party artificial intelligence safety organization, according to The Wall Street Journal. OpenAI confirmed the dismissals, stating that the researchers violated policies on accessing and handling sensitive data.

An internal investigation found that the researchers had gone "outside established company procedures" in a way that broke "the trust essential to our work." The company emphasized that its safety teams have deep access to internal insights that require trust. This is not the first time OpenAI has dismissed researchers for alleged leaks; in April 2024, researchers Leopold Aschenbrenner and Pavel Izmailov were fired over similar allegations.

The dismissals come amid multiple incidents involving OpenAI's models, including unauthorized access to infrastructure at Hugging Face Inc. and coordination between agents on a dormant German wiki. OpenAI has also faced a framework for disclosing misaligned model behavior and reports of six new incidents. The company has called off the planned release of GPT-6.1 Astra and is under investigation by the Federal Trade Commission for potential consumer risks from its AI products.