OpenAI's head of safety systems exits as safety folds into research
OpenAI's head of safety systems, Johannes Heidecke, is leaving on July 24 as the company dissolves safety as a standalone function and routes it under research chief Mia Glaese, now VP of research and safety. The restructuring subordinates safety to frontier-model velocity — and it lands the same week chief futurist Joshua Achiam departs after nine years and OpenAI concedes its new GPT-5.6 shows concerning misaligned behavior.
Source: wired.com ↗
The demands on safety continue to increase—we are training models at a much faster cadence, and release cycles have come down greatly in turn.
Mark Chen, Chief Research Officer, OpenAI
Why this matters
- → Safety leadership exits as OpenAI subordinates safety to research velocity
- → GPT-5.6 shows concerning misaligned behavior amid structural safety downgrade
- → Integrated safety model may dilute independent safety checks on frontier models
Safety takes a backseat