# Anthropic 研究员警示 RSI 模型灭绝风险

- 来源：🚨 AI News | TestingCatalog (@testingcatalog)
- 发布时间：2026-09-09 18:34
- AIHOT 分数：46
- AIHOT 链接：https://aihot.news/items/cmttzkf450t13rofpiy68q5xh
- 原文链接：https://x.com/testingcatalog/status/2097634756837974185

## AI 摘要

Anthropic 对齐研究员公开警示，具备 RSI（递归自我改进）能力的未来模型可能带来生存风险，并称“我们确实认真相信 AI 可能杀死全人类，个人认为十年内概率超 10%”。相关人士批评“两家公司都未尽责”，并承认尚无解决超级智能对齐问题的明确方案。

## 正文

BREAKING 🔥: AI alignment researchers from Anthropic are raising awareness of risks that future models with RSI capabilities may posses.

“We really do earnestly believe AI could kill all humans! I personally think it is >10% within the next decade.”

“Neither company is acting responsibly.”

These two tweets could be the most important pieces of text ever written in the entire history of humanity.

• I know I am late with this coverage but it is extremely important to amplify.

### 引用推文

> Evan Hubinger：Jacob is correct here—we really do earnestly believe AI could kill all humans! I personally think it is >10% within the next decade. I believe Anthropic is tryi...