1
Blind Spots in the Guard: How Domain-Camouflaged Injection Attacks Evade Detection in Multi-Agent LLM Systems
揭秘多智能体LLM系统中域伪装注入攻击如何绕过安全防护，揭示现有检测器的盲区。
arXiv:2605.22001v1 Announce Type: cross Abstract: Injection detectors deployed to protect LLM agents are calibrated on static, template-based payloads…