Gary Marcus 提醒：AI 智能体易受提示词注入攻击
精选理由
用智能体帮你收发邮件前先看看这条：别人藏在邮件里的一句话可能让你的 AI 照做，这不是理论风险。
Gary Marcus 转发 Jeffrey Lee Funk 的警告，指出智能体 AI 需要处理邮件等外部消息，而这些内容不应被直接信任。问题核心在于消息中存在大量隐藏内容，AI 模型无法区分需要阅读的文本和需要执行的指令。该推文目前有 6 次转发和 9 个点赞，链接指向 Futurism 的相关报道。
原文 · Gary Marcus
⚠️ PSA: “AI assistants may fall prey to prompt injection attacks” ⚠️ jeffrey lee funk @jeffreyleefunk Agentic AI requires us to trust our emails and other messages and we really shouldn't. There is too much hidden content and AI models can’t tell the difference between text it’s supposed to read and instructions it’s supposed to follow. futurism.com/artificial-int… 🔗 View Quoted Tweet 💬 0 🔄 6 ❤️ 9 👀 1624 📊 2 ⚡