Another 232papers do not have category tags yet and are not included in filtered results.
提出 Secure-CUA，用动作事务约束计算机使用智能体的不可信影响
Secure-CUA 要限制图形界面里不可信内容对 CUA 决策和 GUI 执行的影响，同时保留合法任务所需的不可信信息。
提出 WebMirage，用局部对抗图像劫持视觉网页 Agent 的浏览器操作
WebMirage 是一个面向视觉定位网络智能体的流水线感知端到端红队攻击框架。
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
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。
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 Aura，用可信内核中介移动 Agent 的身份、感知与动作控制
Aura uses kernel-mediated structured agent interaction, replacing screen-interface mobile assistants..
提出 WAInjectBench 评测网页 Agent 的提示注入检测
WAInjectBench 是网页智能体提示注入检测的基准。作者要看现有检测器能否发现这类攻击：攻击者改网页，使智能体执行目标动作，而不是用户任务。作者称这是首个此类综合基准。
提出 WebInject，用网页像素扰动诱导网页 Agent 执行指定动作
WebInject 是对 MLLM 网页智能体的环境提示注入：修改网页源码，使原始像素上的扰动经过显示器映射后，诱导智能体执行攻击者指定动作。
提出 ARE 框架与图像文本攻击，评测多模态网页 Agent 的对抗鲁棒性
Abstract系统评估多模态网页智能体的脆弱性，指出推理时计算组件被攻破会加剧系统风险。