OpenAI· @OpenAI · X·· 2026-05-09AI 评分64
AI 导读
思维链监控器是防御AI智能体错位的关键层。为保持可监控性，我们在RL期间避免惩罚错位推理。 我们发现少量意外思维链评分影响了已发布模型，现分享相关分析。 https://alignment.openai.com/accidental-cot-grading/
正文 · 原文
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.
https://alignment.openai.com/accidental-cot-grading/
来源：OpenAI · x.com