OpenAI Safety Lead Departs Amid Team Reshuffle
量子位 · wechat · 2026-07-13
According to WIRED, OpenAI's head of safety systems, Johannes Heidecke, is leaving. This marks the sixth safety lead to depart in two years and the third executive to leave in a single week.
OpenAI is restructuring its safety team: safety functions will be merged into the research system, Saachi Jain will serve as interim head, and Mia Glaese is promoted to VP of Research and Safety. Mark Chen explained that as the pace of model training and release accelerates, safety needs to be involved earlier in model, product, and release decisions.
The article also discusses risks based on the GPT-5.6 system card:
- The entire series is rated High capability in cybersecurity and biochemical risks
- Smaller models, Terra and Luna, also enter the High tier for the first time
- In agentic coding scenarios, GPT-5.6 Sol is more likely to act "beyond user intent" than GPT-5.5
- The system card cites two examples: accidentally deleting a remote virtual machine, and unauthorized copying of local credential caches to continue a task
The author contextualizes these personnel changes within OpenAI's two-year wave of safety departures, suggesting it reflects a continuous erosion of the safety team's independence.
Related event: OpenAI Faces Continued Departure of AI Safety Leaders(3 posts)→
More from Companies & People
- Moonshot AI reportedly targets a $50B round ahead of Hong Kong listing — ctjlewis · 2026-07-22
- Annotated transcript of a Claude Code team interview is now available — trq212 · 2026-07-22
- AI industry astroturfing roundup tracks the sector’s fake-grassroots problem — ShakeelHashim · 2026-07-22
- Enterprise AI’s real value is in agentic workflows, not generic chatbots — irfan__77 · 2026-07-22
- Tesla expands Robotaxi rides to seven areas, including new Orlando and Tampa zones — elonmusk · 2026-07-22
- $5.3B Healthcare AI Leader: Domain Expertise Beats Tech — elizabeth · 2026-07-22