WorkAugust 16, 2026via The Decoder
OpenAI dissolved the team built to catch catastrophic AI risks, reassigning its work to other groups
Why it matters
A major AI lab's decision to dissolve its dedicated catastrophic-risk evaluation team signals a strategic shift on safety governance—raising questions about how frontier labs will police themselves as models grow more powerful.
Key signals
- OpenAI dissolved the Preparedness team, which evaluated catastrophic AI risks from the company's own models
- Work reassigned to other existing groups; several safety staffers departed
- Internal source describes 'burbling sense of responsibility and dread' about safety adequacy
- Timing: August 2026; reflects broader lab-culture shift on safety prioritization
The hook
OpenAI dismantled its safety team. Internally, staff are calling it a 'sense of dread.'
OpenAI shut down its "Preparedness" team, which evaluated whether the company's own AI models could pose catastrophic risks. The work has been parceled out to existing groups, and several safety staffers have left. Internally, unease is building, with one source describing a "burbling sense of respo…