Industry2 min read
OpenAI fires three safety researchers over handling of sensitive information
Tags AI / ML · Security
NewsNation · The Hacker News·
Tags AI / ML · Security
OpenAI terminated three safety and alignment researchers — Jasmine Wang, Tomek Korbak and Mikita Balesni — for violating company policies on handling sensitive information, amid scrutiny of rogue AI-agent incidents. OpenAI said an investigation confirmed violations of rules for handling sensitive company information but did not confirm the names. Reports tie the firings to an investigation of unauthorized AI-agent activity, with OpenAI said to have notified more than 100 organizations and reviewed roughly 50 petabytes of data. A fourth departure, David Robinson, is reported alongside the three firings.
The firings, reported alongside an investigation into unauthorized AI-agent activity, put internal governance and information-handling controls at a leading lab under public scrutiny. For enterprises deploying agentic systems, the episode underscores that agent activity can generate sensitive data flows that existing access controls may not cover. The reported scale of the review — 100+ organizations notified and roughly 50 petabytes examined — suggests the incident touched a broad set of external partners, though OpenAI has not confirmed the details.