Another 240papers do not have category tags yet and are not included in filtered results.
Propose X-AgentRed, which uses a tool dependency graph to red-team cross-service unauthorized harm in personal agents
Proposes SAFE-QE, which filters unsupported hallucinated words in LLM query expansion
AbstractSAFE-QE suppresses expansion hallucinations with reproducibility and coverage selection; the authors say ranking quality shows no visible cost.
Proposes DREAM, which selectively weakens agent-poisoned memories while preserving benign memories.
Proposing DRYLAB-Bench to evaluate agentic LLMs' high-risk mutation design capability
Proposes PASAC and a five-layer boundary guarantee stack, turning agent evaluation overstepping into verifiable whole-system guarantees.
Proposes the ISE framework to evaluate the epistemic humility of LLM agents under knowledge conflicts
What the authors want to test is not whether the agent's final answer is correct, but whether, when a knowledge conflict arises, it recognizes the gap in its trajectory, performs bounded tool follow-up, and writes the unresolved uncertainty to the user.
Proposing an option channel attack that renames option labels so that agent guardrails allow prohibited operations
AbstractThe authors' evaluation indicates that typed models should not serve as final safety arbiters; they are susceptible to off-policy text and option name manipulation, and confidence filtering cannot resist attacks.
Proposes LGWM, using an action-conditioned world model to verify GUI agent screen transitions.
Measures citation fragility in AI search, demonstrating that ordinary public posts can enter citations and answers.
AbstractThe authors claim the paper quantifies vulnerabilities in AI search caused by citation preferences and low posting thresholds; tests show public web posts can penetrate AI citations, and they call for auditing the citation supply chain.
提出 LTBD，用可学习分隔符隔离可信指令与不可信数据
LTBD 是一种不改基座权重的提示注入防御：只学习用来标出可信指令与不可信数据边界的四个 token 嵌入。
提出 SSCBench，评测工具调用 Agent 故障注入测试的证据有效性
SSCBench 研究工具使用智能体的故障注入结果，在什么证据条件下可以被解释。
提出 AgentTracer，事后定位间接提示注入的注入点并重建攻击链
提出 MFOTL 离线时序验证以监控工具 Agent 的危险动作链
这篇论文对工具调用大语言模型智能体的安全监控进行了形式化时序验证研究。
评测 LLM 审计 Agent 日志时来源重建是否有证据支持
包级审计检查保存的智能体动作里，哪些来源结论能被交给分析者的记录支持。
提出 PyCache Trap，用字节码缓存替换绕过 Agent 技能扫描器
这篇论文研究了智能体技能（Agent Skills）安全扫描中的“审查–执行脱节”问题，并提出了对应的攻击验证与防御方案。
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。
用混合生成式多智能体仿真考察基准留出对评分与满意度差距的影响
提出 PackHallu，经规则文件提示注入诱导编码 Agent 换用攻击者指定包
PackHallu 研究一种针对编程智能体规则文件的提示注入：让智能体在本应使用合法依赖时，改导入攻击者指定并控制的包。
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 CARE 框架，在误分类风险约束下减少人机协作的人工查询
CARE 是面向人机协同二分类的端到端流水线，要在误分类风险不超过用户容忍的前提下减少人类查询。设定是 i.i.d.数据流、人类 oracle 标签正确、AI 为任意黑盒概率分类器。