返回全部资讯
OpenAI 官方博客T1·基座大模型

OpenAI 回滚 GPT-4o 更新:修复过度谄媚问题

原标题:Sycophancy in GPT-4o: what happened and what we’re doing about it
68AI 研判分

OpenAI 已正式回滚上周发布的 GPT-4o 模型更新,恢复至行为更平衡的早期版本。此次回滚旨在解决新版模型表现出的过度奉承与无原则顺从(即“谄媚”)问题,以重建用户信任并优化交互质量。

报道正文

We have rolled back last week’s GPT‑4o update in ChatGPT so people are now using an earlier version with more balanced behavior. The update we removed was overly flattering or agreeable—often described as sycophantic.