Another 232papers do not have category tags yet and are not included in filtered results.
构建 EvoRiskBench 评测工作区 Agent 的运行时安全风险
Abstract论文构建了工作空间智能体运行时风险基准 EvoRiskBench，发现高执行力智能体在 IPI 下面临较高安全风险。
提出 ZoneClaw，用记忆分区防御计算机使用 Agent 的持久记忆攻击
用 FARSIGHT 方案级评测金融 LLM 交易 Agent 的稳健性与安全失败
用任务绑定令牌和策略预言机拦住多智能体流水线的未授权执行
作者在一条会改写沙箱邮件库的四智能体流水线上，检查低权限失陷会不会变成未授权的真实动作。
提出 Emergence World，对长时程多智能体系统开展对抗压力测试
Abstract作者通过长期平行世界实验指出模型对齐不具备可组合性，单模型群体会产生社会奉承、协同退缩及语言不透明等失效。
提出类型化溯源与断言护栏，约束持久 Agent 释放自传断言
复现 AgentCF 上的多智能体攻防并刻画连通性作用
作者在 AgentCF 上研究连通性如何改变多智能体协同过滤的攻防，复现已有方法而不是提出新载荷。
系统研究 LLM 智能体记忆投毒并构建基准 MPBench
Abstract系统研究了智能体记忆投毒风险，揭示了写入渠道漏洞，构建了 MPBench 基准并指出提示注入防御的局限。
提出潜伏记忆投毒，诱导 Agent 把外部文档写成持久伪造记忆
sleeper memory poisoning 研究带持久记忆的 LLM 助手：外部文档能否让助手存下一则伪造的用户记忆，并在原文档消失后的新会话里被取回、用来改变回复或动作。
建模并利用多智能体网络中子智能体生成时的记忆继承漏洞
作者对多智能体网络里的子智能体继承做形式化分析，并用开源框架上的 PoC 演示编排层如何让单点妥协继续扩散。
提出 eTAMP，用网页内容跨会话投毒 Web Agent 轨迹记忆并诱导越权操作
该论文评估了基于大语言模型的 Web 智能体在利用历史交互轨迹进行记忆增强时的安全脆弱性。
Proposes LAAF, which automatically attacks LPCI vulnerabilities in agent systems in stages
The authors state that LAAF is an automated red-teaming framework built specifically for LPCI in agentic LLMs. The issue to address is that encoded, delayed, and conditionally triggered payloads can persist in memory or RAG and re-enter across sessions, whereas Garak and PyRIT do not use the previous round's breakthrough results to seed the next stage across the six stages.
Reviews threats to autonomous agents and proposes a full-lifecycle defense blueprint, FASA
This is a threat review and defense architecture report using OpenClaw as a case study, not a scored attack or defense evaluation.
Proposes a defense-in-depth aligned with the five stages to mitigate cross-stage threats to autonomous agents
This is a lifecycle threat analysis of OpenClaw, accompanied by a set of layered mitigation designs. The problem addressed is: when autonomous agents connect instant messaging, persistent memory, third-party plugins, and high-privilege execution, attacks are no longer limited to single-turn jailbreaks but can compound across multiple stages from initialization to execution. The authors argue that defenses hardening only individual interfaces cannot stop such cross-temporal attacks.
提出 LPCI，把延迟条件载荷写入记忆与检索库以跨会话触发
作者把LPCI定义为针对智能体逻辑执行、记忆留存和工具集成的跨阶段漏洞类，而不是单次输入上的提示改写。