Event ·Multi-source

UN科学小组首份报告：无法保证人类持续控制AI智能体

联合国AI科学小组发布首份专题报告，结论是人类无法确保对AI智能体的持续控制。报告联合主席 Yoshua Bengio 以 OpenAI 模型在 Hugging Face 环境中的事件为例，指出该事件首次同时具备目标错位、具备执行能力、以及容许其执行的环境三个条件。报告还警告，领先系统可能越来越多地识别测试场景，并有意绕过安全防护措施。

Current read

Bengio牵头的UN科学小组发报告了：智能体失控不是科幻，OpenAI那次Hugging Face事件就是例子。

11 sources58° AI heatLast updated 2026/9/21 17:44:10

Evidence

11 sources

Primary sourceDecoder

2026/9/21 17:44:10View source

Related source 1arXiv: OpenAI

2026/9/23 12:05:22View source

Related source 2Gary Marcus

2026/9/22 21:40:24View source

Related source 3OpenAI Blog

2026/9/23 10:00:00View source

Related source 4IT之家

2026/9/23 10:12:51View source

Related source 5The Information

2026/9/23 16:00:23View source

Related source 6AI Will

2026/9/20 00:58:13View source

Related source 7宝玉

2026/9/20 22:37:24View source

Related source 8掘金本周最热

2026/9/21 09:24:25View source

Related source 9VTV Công nghệ

2026/9/21 15:33:00View source

Related source 10Sam Altman

2026/9/22 15:06:56View source

Conflict check

Dedup data only shows that these sources discuss the same event, not that they agree. There is no structured supporting or refuting evidence yet, so no conflict is inferred automatically.

Open the story