OpenAI News·· 14 h ago
How we detect frontier model misalignment
How we detect frontier model misalignment
AI summary
OpenAI uses layered monitoring to detect frontier model misalignment and unauthorized agent behavior so responders can intervene before incidents escalate.
Selection record
Threshold 60Official, first-handFirst 42Second 38
Not admittedSum of both 80 < twice the threshold 120
- Source tier
- Official, first-hand; this tier's threshold is 60
- Pre-filter
- passed：OpenAI前沿模型失准检测与Agent监控
A model scores each item twice, independently, against one written standard, out of 100. An item is admitted only when the two scores add up to twice the threshold. The threshold is set per source tier.
Source: OpenAI News · openai.com