[论文学习]LLM 代理的隐私黑洞:外部存储个人数据的提示注入攻击基准测试深度解读
Benchmarking Prompt-Injection Attacks on Tool-Integrated LLM Agents with Externally Stored Personal Data (OpenReview ~2026)
摘要
本文深度解析了 2026 年发表于 OpenReview 的论文《Benchmarking Prompt-Injection Attacks on Tool-Integrated LLM Agents with Exte…
2026/7/2 0:15:55