Skip to content
Trending storyWatching

AI 拒答能力被高估

1 reports1 sourcesUpdated 2 days ago

What you need to know

AI summary

MIT Technology Review 发文指出，业界把 AI 的“拒绝回答”当成了安全体系里的承重墙，这种信任可能过高。文章认为，一旦模型在特定情境下不再拒绝，原本依赖拒答建立的安全假设就会失效。 该观点来自 MIT Technology Review 的评论，并非新披露的技术结论或实验结果。

Generated by AI from the reporting · updated 2 days ago

Latest updateOct 9 09:00
We're putting too much faith in AI's ability to say no

Timeline

Follow the reports to see every side of the story.

Oct 9
  1. MIT Technology Review · AI
    We're putting too much faith in AI's ability to say no

    MIT Technology Review argues that refusal has become the "load-bearing wall" of AI safety.

Heat trend for this story

Not enough continuous data yet to chart a trend.