OpenAI分析意外思维链评分对模型影响

作者：OpenAI / @OpenAI
发布时间：2026-05-08T20:19:04.000Z

Chain of thought monitors are a key layer of defense against AI agent misalignment. To preserve monitorability, we avoid penalizing misaligned reasoning during RL.

We found a limited amount of accidental CoT grading which affected released models, and are sharing our analysis.

链接卡片：alignment.openai.com We found limited accidental CoT grading in some released models, fixed the affected reward pathways, and found no clear evidence that monitorability degraded.