AI 热点

FrontierBreach 基准揭示企业 AI 智能体工作流的攻击成功率差异

PromptArmor更新于 11月19日 08:40

Curiously, models from China have a higher ASR than models from the US on average, which was disproportionately due to their performance against arbitrary code

AI 导读

PromptArmor 发布 FrontierBreach,评测企业 AI 智能体工作流中的间接提示词注入攻击,九个行业的平均攻击成功率为 34.9%,医疗与社会援助工作流达到 44.4%。中国模型的平均 ASR 高于美国模型,差异主要来自任意代码执行、网络钓鱼和数据破坏任务;它们在数据窃取、勒索软件和持久化攻击中也表现出更高的攻击成功率。

推荐理由

精选AI 评分 68

该基准按行业和攻击类型拆解智能体工作流风险,可用于比较不同模型在间接提示词注入下的安全差异。

主题

标签

正文

Curiously, models from China have a higher ASR than models from the US on average, which was disproportionately due to their performance against arbitrary code execution, phishing, and data destruction tasks.

These models performed better than their US counterparts on Data Theft, Ransomware, and Persistence attacks.

来源

PromptArmor