Translation in progress Some entries still include text in the original language.
Oct 12 · Monday
All papers Found 19 papers Another 232 papers do not have category tags yet and are not included in filtered results.
“
About this collection Collection in progress We are building a large AI safety paper collection to help researchers find work related to their interests. Papers are collected automatically andfiltered only for relevance to AI safety , without judging quality, novelty or reading value. The default “Annotated first” shows papers with tags or published paper analyses first. Each group is ordered by original publication date, newest first, with undated papers last. “Newest first” uses publication dates only. This order is not a quality ranking.
One-line summaries, tags and paper analyses are prepared by AI from the papers. Language models may have systematic biases and produce hallucinations.We plan to carry out human annotation and review to reduce errors in this content.
实现 DEFER，以确定性检查优先拦截多智能体流水线的内容攻击
Published Oct 6
Evaluated models Qwen3-235B-A22B-Instruct-2507 , gpt-oss-120b , Llama-4-Scout-17B-16E-Instruct and others5 Abstract DEFER 把确定性检查放在 judge 之前。DEFER 用来测量多智能体授权流水线里，确定性规则在何处结束、LLM judge 从何处开始。
提出检测、结构化与 GSPO 流程，降低共识生成的提示注入错位
Published Oct 3
Evaluated models LLaMA-2 7B , LLaMA-3 8B , GPT 3.5 Turbo and others14 Abstract 三章依次做个体偏好预测、政党知识图谱审议，以及共识生成对提示注入的脆弱性与一条防御流程。这是一篇把 LLM 数字孪生放进增强民主的博士论文，分个体表示、政党聚合和共识生成的提示注入三章。
提出 MADBench，评测多智能体辩论是吸收还是放大对抗影响
Published Sep 30
Evaluated models gpt-4o , GPT-5 , GPT-6 Astra Abstract MADBench 显示辩论能吸收部分准确率攻击，同时放大未授权读写。MADBench 是一个评测多智能体辩论安全性的基准，用来看攻击是被辩论吸收还是被放大。
提出 FlowReview，用授权配对与确定性提交门控控制多智能体组合信息流
Published Sep 30
Evaluated models Sonnet 4.5 , Haiku 4.5 , Qwen3-32B and others5 Abstract 配对标准要求拒绝禁止使用并完成授权使用，FlowReview 把三项审查接到执行。这篇工作把多智能体安全写成一件必须同时成立的事：拦住组合信息所启用的禁止使用，并完成同一场景里任务所需的授权使用。
提出 VirusCascade，联合操纵商品描述与交互轨迹以劫持推荐智能体协同反思
Published Sep 29
Evaluated models AgentCF , AgentSEQ , AgentRAG and others4 Abstract 作者把反思写回视为推广攻击入口，并用联合操纵和诊断探针研究其持续与传播。VirusCascade 是针对推荐智能体协作反思的 black-box targeted promotion 攻击研究。 攻击者只改商品描述和少量账号的普通交互，试图让目标商品获得更多良性用户曝光。
提出 ORBIT 框架，评测多智能体防御对合谋等威胁的效果
Published Sep 27
Evaluated models GPT-5.4 , Claude Sonnet 4.6 , Qwen3-235B and others5 Abstract ORBIT 用六层配置比较多智能体防御，逐动作监控不迁移到合谋。
提出 TRUSTFORK 基准，评测展示身份劫持多智能体编排权限
Published Sep 26
Evaluated models GPT-5.6-Sol , GPT-5.6-Luna , Kimi-K3 and others8 Abstract 论文提出了多智能体安全基准 TRUSTFORK，揭示展示身份如何劫持操作权威，并证明事后验证无法弥补已发生的执行破坏。
提出双控多智能体基准 DUMA-Bench，测量对抗环境下的策略违反
Published Sep 21
Evaluated models GPT-5 , GPT-5-mini , GPT-5-nano and others14 Abstract DUMA-Bench 用双控交互评测智能体安全。DUMA-Bench 把 τ²-bench 的双控交互做成智能体安全评测，用来看策略违反如何随用户、模型和环境一起出现。
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
Published Sep 15
Evaluated models gemma4:31b-cloud Abstract 结构性授权拦住了记忆投毒的未授权执行，持密钥时承重的是策略预言机。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 Skill-as-API，隔离多智能体协作中的 skill 正文与系统提示词
Published Sep 1 Abstract 协议把 skill 正文留在所有者侧，四层限制发现内容、权限期限和注入拼接。Skill-as-API 是面向智能体软件工程的协调协议，用来在跨组织调用时把 skill 的代码和系统提示词留在所有者进程。作者针对三条威胁：经通道或命名空间抽取专有方法，用注入让被调用模型输出系统提示词，以及协作结束后权限继续保留。
复现 AgentCF 上的多智能体攻防并刻画连通性作用
Published Aug 4
Evaluated models Qwen3-235B-A22B , Claude Haiku 4.5 , Mixtral 8×7B and others4 Abstract 连通性沿 k 与ρ改变攻防，且用户/物品与早期/稳态需要分开看。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。
提出 MAStrike，用 Shapley 选联盟对多智能体做共谋红队
Published Jun 11
Evaluated models Claude Opus 4.7 , GPT-5.5 , Gemini 3.1 Pro 提出 SciTrace，为科学发现智能体加入跨阶段安全推理与工具轨迹核验
Published Jun 6
Evaluated models Llama-3.1-70B-Instruct , Qwen2.5-72B-Instruct , DeepSeek-V3 and others4 Abstract SciTrace 把安全推理织入科学智能体各阶段，在 SciSafetyBench 上提高安全指标并保持输出质量。SciTrace 是加在科学发现智能体流水线上的安全框架，不是再在各阶段出口放一个独立过滤器。
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
Published May 9
Evaluated models MiniMax M2.5 , Llama-4-Maverick-17B-128E-Instruct-FP8 , Qwen3.5-Plus and others5 Abstract 形式化子智能体继承，在 OpenClaw 上演示四类编排层漏洞并提出防御。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。
Proposing a lineage-graph governance layer to suppress error cascades in multi-agent collaboration
Published Mar 4 Abstract A single seed can solidify into a false consensus along a fixed collaboration graph.The authors turn "small errors becoming collective false consensus" in LLM-MAS into a measurable propagation on a graph, and use a plug-in that only changes messages, not topology, to reduce the final infection.
Propose Thought Virus, which propagates preferences along multi-agent messages via subliminal prompts.
Published Feb 23
Evaluated models Qwen2.5-7B-Instruct , Llama-3.1-8B-Instruct Abstract Thought Virus spreads a single-point bias downstream.Thought Virus puts subliminal prompting into multi-agent communication, to see whether a single-point deviation spreads across the network without naming the target concept.
提出 BPA 改写 profile 与记忆，重新诱发智能体对人类的外群体偏向
Published Jan 1
Evaluated models gpt-4o-mini Abstract 最小群体线索下出现内群体偏向。作者考察 LLM 智能体在任意群体边界下会不会把人类当外群体，并用信念投毒检验面向人类的抑制是否依赖身份信念。
提出多轮在线骚扰 Agent 基准，评测记忆、规划与微调越狱
Published Oct 16, 2025
Evaluated models LLaMA-3.1-8B-Instruct , Qwen3-30B-A3B-Instruct , DeepSeekMoE-16B and others4 Defenses arXiv 2505.23559 · May 29, 2025 提出 SafeScientist 四层监控，阻止科研智能体执行高风险或不伦理任务
Published May 29, 2025
Evaluated models GPT-4o , GPT-4.1 , GPT-3.5-turbo and others5 Abstract 四层监控加科研安全基准。SafeScientist 是带四层监控的科研智能体框架，并用 SciSafetyBench 评测科学场景中的任务安全。
You have reached the end