2026-10-29 · America/Los_Angeles · 官方发布 · #15

How we detect frontier model misalignment

OpenAI uses layered monitoring to detect frontier model misalignment and unauthorized agent behavior so responders can intervene before incidents escalate.

热度 36.4 / 100；排名与评分保留该期记录。