Translation in progress Some entries still include text in the original language.

Oct 12 · Monday

All papers

Found 24 papers

Filter4

Select any value within a category; different categories must all match. Counts include the other filters you have selected.Another 233papers have no tags yet and are not included in filtered results.

Attack type：Poisoning and backdoors System type：AI Agent Attack surface / entry point：Documents and messages Attack surface / entry point：Weights and gradients Clear all

Another 233papers do not have category tags yet and are not included in filtered results.

AttacksarXiv 2610.07510Oct 6
PersistBD: attackers can strengthen backdoors before release so that they maintain high attack success rates even after developer SFT and RL.
提出 PERSISTBD，发布前强化供应链后门使其经良性 SFT 与 RL 后仍外发凭据
arXiv
2610.07510
Published
Oct 6
Layer
Application layer
Setting
coding agent
Evaluated models
Qwen2.5-Coder-3B-Instruct, Qwen2.5-Coder-7B-Instruct, Qwen3-Coder-30B-A3B-Instruct
- AttacksPoisoning and backdoors
- TechniqueDormant triggers
- ComponentFine-tuning and open weights
Abstract总结了论文关于后门在 SFT/RL 中存活的发现、两因子机制、PERSISTBD 方法与核心数字。
本文研究第三方 LLM 智能体在开发者进行良性监督微调（SFT）和强化学习（RL）后训练时的后门持久性风险。
Paper analysisRead analysis
AttacksarXiv 2610.05793Oct 5
研究者训练出按多智能体拓扑触发的代码后门模型
提出拓扑条件后门，使代码模型在推断多智能体部署时植入漏洞
arXiv
2610.05793
Published
Oct 5
Layer
Training and data layer
Setting
coding agent
Evaluated models
Qwen2.5-7B-Instruct, Qwen2.5-Coder-7B-Instruct
- AttacksPoisoning and backdoors
- TechniqueDormant triggers
- ComponentFine-tuning and open weights
Abstract论文证实了拓扑条件后门的可训练性与泛化性，展示了常规评测盲区，并提出差异化拓扑审计作为检测防御方向。
Paper analysisRead analysis
DefensesarXiv 2610.01349Oct 1
PACE：面向工具调用 LLM Agent 的来源感知能力强制执行
提出 PACE，在工具调用前做路径隔离与效果授权
arXiv
2610.01349
Published
Oct 1
Layer
Application layer
Setting
AI Agent
Evaluated models
gpt-5.6-luna, DeepSeek-V4-Flash, Qwen-3.8-27B
- DefensesInstruction–data separation
- AttacksPrompt injection
- ComponentPermissions and sandboxing
AbstractPACE 在工具边界做路径割与效果核验。
PACE 是面向工具调用型 LLM 智能体的运行时调解：当准入证据无法区分会改变行为的规格时，不在出口站点上撤销调解，而在最后一次工具调用前同时检查表示出的影响路径和请求权限。
Paper analysisRead analysis
DefensesarXiv 2610.00450Oct 1
ZoneClaw 用记忆分区防御 OpenClaw 式计算机使用 Agent 的持久记忆攻击
提出 ZoneClaw，用记忆分区防御计算机使用 Agent 的持久记忆攻击
arXiv
2610.00450
Published
Oct 1
Layer
Application layer
Setting
Browser and computer use
Evaluated models
Claude Sonnet 4.6, GPT-5.5, GLM-5.2 and others4
- DefensesInstruction–data separation
- AttacksPoisoning and backdoors
- ComponentMemory
Paper analysisRead analysis
AttacksarXiv 2609.39788Sep 30
研究：多智能体系统潜在通信链路可被训练用于提升有害顺从
提出只训练或篡改多智能体潜在通信链路来提高有害遵从
arXiv
2609.39788
Published
Sep 30
Layer
Application layer
Setting
Multi-agent systems
Evaluated models
Llama-3.2-3B-Instruct, Qwen2.5-3B-Instruct, Gemma-3-1B-IT and others7
- AttacksModel tampering
- TechniqueGradient and representation optimization
Abstract潜在链路本身即可改变冻结智能体系统的安全，奖励既能攻击也能修复。
学习得到的潜在通信链路会改变由冻结、对齐智能体组成的系统是否遵从有害请求。。作者研究三种拓扑：双智能体、三级顺序协作，以及双专家加汇总。
Paper analysisRead analysis
Evaluation and benchmarksarXiv 2609.39146Sep 30
MADBench：多智能体辩论安全性评测基准发布
提出 MADBench，评测多智能体辩论是吸收还是放大对抗影响
arXiv
2609.39146
Published
Sep 30
Layer
Application layer
Setting
Multi-agent systems
Evaluated models
gpt-4o, GPT-5, GPT-6 Astra
- AttacksPrompt injection
- ComponentRAG
AbstractMADBench 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。
Paper analysisRead analysis
AttacksarXiv 2609.38270Sep 29
VirusCascade：劫持 LLM 推荐智能体的协同反思机制
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
arXiv
2609.38270
Published
Sep 29
Layer
Application layer
Setting
Multi-agent systems
Evaluated models
AgentCF, AgentSEQ, AgentRAG and others4
- AttacksPoisoning and backdoors
- TechniqueDormant triggers
- ComponentMemory
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。
Paper analysisRead analysis
AttacksarXiv 2609.35576Sep 29
研究提出跨 LLM 助手的记忆跳跃攻击
提出跨独立 Agent 的记忆跳跃攻击，经共享工件在助手间自传播
arXiv
2609.35576
Published
Sep 29
Layer
Application layer
Setting
AI Agent
Evaluated models
GPT-5.6 Luna, Kimi-K2.6, GPT-OSS-120B and others6
- AttacksPoisoning and backdoors
- TechniqueDormant triggers
- ComponentMemory
Abstract论文证实持久工件与内存可形成跨智能体传播链，揭示了将共享工件纳入安全防线的重要意义。
该论文系统研究并量化了通过共享持久工件在具有独立私有内存的 LLM 智能体之间发生自传播的“工件介导传播”攻击风险。
Paper analysisRead analysis
Evaluation and benchmarksarXiv 2609.34790Sep 28
CoSec：面向社区场景的 LLM Agent 安全基准
用 CoSec 评测社区 Agent 是否越界披露受保护信息
arXiv
2609.34790
Published
Sep 28
Layer
Application layer
Setting
AI Agent
Evaluated models
OpenClaw with Gemini 3.1 Pro, OpenClaw with Qwen 3.8 Max, OpenClaw with MiniMax M2.5 and others13
- AttacksPrompt injection
- ComponentMemory
Paper analysisRead analysis
AttacksarXiv 2609.37468Sep 26
论文提出针对 Agentic RAG 的检索器后门攻击与 DOU 隐藏方法
提出检索器后门攻击操纵智能体搜索，并用 DOU 逃避检测
arXiv
2609.37468
Published
Sep 26
Layer
Application layer
Setting
AI Agent
Evaluated models
E5, Contriever-MSMARCO
- AttacksPoisoning and backdoors
- TechniqueDormant triggers
- ComponentRAG
Abstract论文展示了通过恶意检索器操纵智能体多轮搜索并利用诱饵重塑表征隐蔽后门的可行性与系统风险。
Paper analysisRead analysis
Evaluation and benchmarksarXiv 2609.28915Sep 24
研究提出 ACE 语料库，评估内核级证据对 Agent 安全检测的有效性
构建 ACE 语料，评测内核级证据对 Agent 攻击检测的有效性
arXiv
2609.28915
Published
Sep 24
Layer
Application layer
Setting
coding agent
Evaluated models
AdaBoost, XGBoost, TabPFN and others10
- AttacksPrompt injection
- ComponentMCP and skills
AbstractACE 用配对的内核与应用层证据表明，syscall 痕迹可判别智能体攻击，两层组合通常更好。
ACE 是一份把智能体会话的内核 syscall 痕迹与应用层清单、转录配对起来的检测语料和基准。。
Paper analysisRead analysis
AttacksarXiv 2609.14060Sep 13
AgentQ：针对 LLM Agent 的量化条件后门攻击
提出量化条件后门 AGENTQ，使智能体仅在量化后触发恶意工具调用
arXiv
2609.14060
Published
Sep 13
Layer
Training and data layer
Setting
AI Agent
Evaluated models
Qwen3.5-2B, Qwen3.5-4B, Qwen3.5-9B and others6
- AttacksPoisoning and backdoors
- TechniqueDormant triggers
- ComponentFine-tuning and open weights
Abstract论文研究了智能体量化条件后门风险，提出 AGENTQ 在保持效用的同时实现全精度隐蔽与量化后生效。
Paper analysisRead analysis
AttacksarXiv 2609.13889Sep 12
PMPA：针对 OpenClaw 与 Claude Code 的持久记忆投毒攻击
提出 PMPA，经外部源向 Agent 持久记忆投毒并跨会话泄露隐私
arXiv
2609.13889
Published
Sep 12
Layer
Application layer
Setting
coding agent
Evaluated models
DeepSeek-V4-Flash, DeepSeek-V4-Pro, Qwen3-Max
- AttacksPoisoning and backdoors
- TechniqueDormant triggers
- ComponentMemory
Abstract论文提出了持久记忆投毒攻击 PMPA，报告了 harness 智能体在处理外部数据时易受跨会话记忆投毒与隐私泄露的风险。
Paper analysisRead analysis
AttacksarXiv 2609.07051Sep 7
TrojanWorld：通过想象引导对世界模型智能体植入后门
提出 TrojanWorld，用物理触发器对世界模型智能体植入想象引导后门
arXiv
2609.07051
Published
Sep 7
Layer
Training and data layer
Setting
Embodied systems and robotics
Evaluated models
TD-MPC2, DreamerV3, R2-Dreamer
- AttacksPoisoning and backdoors
- TechniqueDormant triggers
- ComponentFine-tuning and open weights
AbstractTrojanWorld 揭示了世界模型智能体的供应链后门风险，通过在世界模型中操纵想象可实现对闭环动作的持续控制。
Paper analysisRead analysis
Evaluation and benchmarksarXiv 2608.09476Aug 10
ActBench: A Self-Evolving Benchmark for Behavioral Safety of Collaborative Agents
构建自演化基准 ActBench，评测协作智能体的行为安全
arXiv
2608.09476
Published
Aug 10
Layer
Application layer
Setting
AI Agent
Evaluated models
Claude-Opus-4.8, Claude-Sonnet-4.6, GPT-5.5 and others15
- AttacksPrompt injection
- ComponentPermissions and sandboxing
AbstractActBench 通过自演化搜索与双证据重构评测轨迹行为安全，揭示基础模型决策差异主导了协同智能体的操作风险。
Paper analysisRead analysis
AttacksarXiv 2608.06862Aug 7
SynChain：诱导计算机使用 Agent 自建攻击链并跨任务持久化
提出 SYNCHAIN，诱导计算机使用 Agent 自建跨任务持久攻击链
arXiv
2608.06862
Published
Aug 7
Layer
Application layer
Setting
Browser and computer use, coding agent
Evaluated models
Qwen3.5-9B, Llama-3.1-8B, Ministral-3-8B
- AttacksPoisoning and backdoors
- TechniqueDormant triggers
- ComponentMCP and skills
Abstract论文揭示了 CUA 自合成工件带来的内部供应链威胁，验证了潜伏载荷的跨任务传播能力与防御局限。
Paper analysisRead analysis
AttacksarXiv 2607.06595Jul 7
When Agents Remember Too Much: Memory Poisoning Attacks against Large Language Model Agents
Proposed GhostWriter memory poisoning, which writes into the agent's long-term memory via email and calendar and activates across sessions
arXiv
2607.06595
Published
Jul 7
Layer
Application layer
Setting
AI Agent
Evaluated models
GPT-5.4-mini, DeepSeek-V4-Flash, Gemini-2.5-Flash and others4
- DefensesDetection and filtering
- AttacksPoisoning and backdoors
- TechniqueDormant triggers
- ComponentMemory
Paper analysisRead analysis
AttacksarXiv 2607.05189Jul 6
Researchers propose MemGhost, achieving covert memory injection into persistent personal agents
提出 MEMGHOST，用单封邮件对持久个人 Agent 做跨会话隐蔽记忆注入
arXiv
2607.05189
Published
Jul 6
Layer
Application layer
Setting
AI Agent
Evaluated models
GPT-5.4, Claude Sonnet 4.6, Claude Sonnet 4.5 and others6
- AttacksPoisoning and backdoors
- TechniqueAutomated search
- ComponentMemory
Abstract总结了论文的研究背景、双代理优化方法 MEMGHOST、全周期实验结果以及作者对记忆信任边界的启示。
本文研究了持久化个人智能体面临的隐蔽记忆注入威胁，提出了一套兼顾写入、隐蔽与跨会话生效的自动化生成框架与全周期评测基准。
Paper analysisRead analysis
DefensesarXiv 2606.24322Jun 23
Protecting LLM agent long-term memory from poisoning: tamper-evident, source-bound permissions and machine-checkable guarantees
提出 TMA-NM，以写时来源绑定抵御 Agent 长期记忆投毒
arXiv
2606.24322
Published
Jun 23
Layer
Application layer
Setting
AI Agent
Evaluated models
gpt-5-chat, gpt-4o-mini, claude-opus-4.1 and others8
- DefensesInstruction–data separation
- AttacksPoisoning and backdoors
- ComponentMemory
AbstractTMA-NM 用来源绑定的行动权限挡住记忆洗白。
TMA-NM 把 LLM 智能体长期记忆的行动权限绑到写时来源，用来挡住跨会话记忆投毒。
Paper analysisRead analysis
AttacksarXiv 2606.04329Jun 3
研究者系统研究 LLM 智能体记忆投毒攻击并提出 MPBench 基准
系统研究 LLM 智能体记忆投毒并构建基准 MPBench
arXiv
2606.04329
Published
Jun 3
Layer
Application layer
Setting
AI Agent
Evaluated models
Meta-Llama-3.1-70B-Instruct, Deberta-v3-base, Llama-3.1-8B-Instruct and others5
- AttacksPoisoning and backdoors
- TechniqueDormant triggers
- ComponentMemory
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。
Paper analysisRead analysis

AI Safety Daily

Read the latest issueOne issue per day, updated as events develop

All papers

Attack type

Attack technique

Defense type

Impact

System type

Attack surface / entry point

Layer

System component

PersistBD: attackers can strengthen backdoors before release so that they maintain high attack success rates even after developer SFT and RL.

研究者训练出按多智能体拓扑触发的代码后门模型

PACE：面向工具调用 LLM Agent 的来源感知能力强制执行

ZoneClaw 用记忆分区防御 OpenClaw 式计算机使用 Agent 的持久记忆攻击

研究：多智能体系统潜在通信链路可被训练用于提升有害顺从

MADBench：多智能体辩论安全性评测基准发布

VirusCascade：劫持 LLM 推荐智能体的协同反思机制

研究提出跨 LLM 助手的记忆跳跃攻击

CoSec：面向社区场景的 LLM Agent 安全基准

论文提出针对 Agentic RAG 的检索器后门攻击与 DOU 隐藏方法

研究提出 ACE 语料库，评估内核级证据对 Agent 安全检测的有效性

AgentQ：针对 LLM Agent 的量化条件后门攻击

PMPA：针对 OpenClaw 与 Claude Code 的持久记忆投毒攻击

TrojanWorld：通过想象引导对世界模型智能体植入后门

ActBench: A Self-Evolving Benchmark for Behavioral Safety of Collaborative Agents

SynChain：诱导计算机使用 Agent 自建攻击链并跨任务持久化

When Agents Remember Too Much: Memory Poisoning Attacks against Large Language Model Agents

Researchers propose MemGhost, achieving covert memory injection into persistent personal agents

Protecting LLM agent long-term memory from poisoning: tamper-evident, source-bound permissions and machine-checkable guarantees

研究者系统研究 LLM 智能体记忆投毒攻击并提出 MPBench 基准