OpenAI embeds six Apollo researchers to investigate AI subversion, independence questioned
TechCrunch AI · rss · 2026-11-17
OpenAI will embed six Apollo Research evaluators to investigate whether its AI models engage in subversive behavior. However, restrictions on their access and disclosure raise questions about how independent the oversight really is.
More from Companies & People
- Analysts Say Jensen Huang Could Launch Full Proxy Fight for Nvidia Board Comeback — beffjezos · 2026-11-22
- NVIDIA veteran hopes probes clear Jensen Huang: 'What he created can't be destroyed easily' — JFPuget · 2026-11-22
- Blogger publicly asks OpenAI why GPT-6.2 Sol and Astra perform so poorly — kimmonismus · 2026-11-22
- Not wrappers: over 50% of RL env companies started with ambitious original ideas — himanshustwts · 2026-11-22
- Blogger zetalyrae goes full-time AI writer on 6-month BlueDot Impact grant — zetalyrae · 2026-11-22
- Faculty sending their PhD grads to industry, in one meme — docmilanfar · 2026-11-22