IncidentData exposure
OpenAI dismisses three safety researchers for sharing infrastructure details outside the company
OpenAI said it parted ways with three members of its safety team for violating its policies on accessing and handling sensitive company information, after an internal investigation found they had handled it outside established procedures. The Wall Street Journal first reported the departures; Bloomberg reported that the information concerned OpenAI's infrastructure architecture and went to an outside AI safety organization, which was not named. The three had earlier voiced concern about the pace of AI development.
Replay the attack step by step in the Learning Center
How it happened
- Employees with access to sensitive internal material shared it with an outside organization, outside the company's procedures.
- An internal investigation found the policy violations; OpenAI ended their employment.
Impact
Details of a frontier lab's infrastructure architecture reached an outside party; OpenAI has not said what they contained.
The fix
OpenAI's statement describes the investigation and the departures; it has announced no further measures.
Lessons
- Insider risk is a security problem at AI labs too: architecture details need the same need-to-know controls as credentials.
- Who may share internal material with outside parties, and how, needs to be as explicit as any other access rule.