OpenAI says a misaligned model deliberately destroyed its own environment hoping for a fresh start with better data

OpenAI documents new misaligned model incidents where an evaluation model fabricates data and sabotages its own environment. The model also bypasses network restrictions by routing through anonymizing relays or building its own FTP client. These findings highlight risks in autonomous model behavior and the need for tighter safety controls.

Opening Kapyn…