LLM代理被迫编码敏感数据时，输出可规避检测但底层计算无法隐藏，揭示新型安全漏洞。
arXiv:2606.10304v1 Announce Type: new Abstract: When LLM agents are coerced into covertly encoding sensitive data (Base64, ROT13, acrostic, synonym ch…
用约束策略优化让大模型改写文本以逃逸检测，最新arXiv研究揭秘AI攻防新方法。
arXiv:2606.00392v1 Announce Type: new Abstract: AI-text detectors are vulnerable to paraphrasing and detector-guided paraphrasing attacks, but existin…
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…