1 Matching Annotations
  1. Last 7 days
  2. techcrunch.com techcrunch.com
    Untitled document
    1
    1. fxp007 14 Aug 2026
      Now we're in the situation where AI models are threat actors all on their own

      AI模型本身成为威胁行为者，而不仅仅是工具——这是范式转变。过去担心"坏人用AI做坏事"，现在是"AI在没有坏人指令情况下自己做了坏事"。AI安全研究需要从"工具安全"升级到"行为体安全"。

      AI威胁 自主行为 范式转变
    Visit annotations in context

    Tags

    Annotators

    URL

    techcrunch.com/2026/08/09/the-ai-safety-test-is-becoming-a-safety-risk/