Three fired OpenAI safety researchers deny leaking secrets

Three OpenAI safety researchers who were dismissed for alleged leaks have stated their side of the story, according to a Wall Street Journal report cited by IT Home on October 25, 2026. The three, Tomek Korbak, Mikita Balesni and Jasmine Wang, deny leaking confidential information and question the official rationale for their dismissal, suggesting the matter may be connected to their long-standing advocacy for greater transparency in AI systems. All three had previously published views on AI risks.

The Wall Street Journal report outlines how the conflict unfolded. OpenAI's safety team suspected the three had shared internal model architecture details with an external safety researcher and asked them to unlock their personal phones. Investigators then found Signal chat records with disappearing messages enabled, and one week later all three were fired simultaneously.

OpenAI said the dismissals followed an internal investigation that concluded the three had violated sensitive information policies, adding that the firings were unrelated to the safety concerns they had raised. The researchers pushed back, stating the company never told them in writing which specific information had been improperly handled, and expressing concern the episode could discourage independent safety auditors from working with AI companies. The situation now stands at a standoff: the company maintains the dismissals were justified, while the three researchers deny leaking and dispute the official account, with no resolution reported so far.