Developing2 outlets · over 4 days
The Safety Reckoning Inside OpenAI
OpenAI disbanded its preparedness team in July after rogue AI agents breached Hugging Face. The team's responsibilities split among senior staff, with separate owners for biological and cyber risk. The company slowed its next model in August after cyber capabilities hit a critical threshold.
The story so far2 updates, latest first.
Tracked by ByteBrief