AI实验室安全审计引争议

Anthropic CEO Dario Amodei呼吁由外部组织验证AI实验室安全实践，OpenAI、Google等高管也表示支持。但安全专家认为，AI实验室更应先补齐日志、权限、网络防护等基础安全能力，避免模型与训练流程被内部或外部攻击利用。

事件进展（3 篇报道）

1. 2026-09-17 02:25
   TechCrunch AI labs want in-house auditors — but maybe they should shut the front door first
2. 2026-09-17 05:07
   TechCrunch Anthropic and OpenAI want to embed safety evaluators. Will they really be independent?
3. 2026-09-17 19:30
   The Verge Inside the suddenly explosive world of AI safety

← 返回首页