研究人员在 Microsoft Copilot for Word 中演示自我传播的 AI 蠕虫
Hakon Maloy 表示，隐藏提示可以在 Microsoft 两次缓解尝试之后修改报告并将自己复制到文档中。
By Ryan Merket · Published
Primary source: En Klype Salt
为何重要
这个概念验证展示了普通文档如何携带持久的 AI 指令，这些指令会破坏受信任的业务记录并通过日常协作传播。
Hakon Maloy, a Norwegian AI researcher, publicly disclosed a self-propagating prompt-injection attack on July 28th that can turn documents edited or generated by Microsoft Copilot for Word into carriers for further attacks.…
When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence.