Skip to content

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

Oct 12 · Monday

All papers

Found 33 papers
Filter3

Attack type

Prompt injection13Poisoning and backdoors20

Attack technique

Gradient and representation optimization4Automated search5Encoding and obfuscation3Dormant triggers3

Defense type

Detection and filtering4

Impact

Unauthorized actions6Data and privacy leaks10Output manipulation33

System type

Models and APIs7LLM applications20AI Agent6

Attack surface / entry point

User input57Web pages and interfaces16Documents and messages27Tool responses16Tool and skill descriptions5Code repositories and dependencies6Retrieval corpus26Memory writes7Inter-agent messages9Images, audio and video50Training data32Weights and gradients15Distribution channels16Inference services and hardware7

Layer

Model layer4Application layer26

System component

RAG20Guardrails and judges3Embeddings3
Select any value within a category; different categories must all match. Counts include the other filters you have selected.Another 265papers have no tags yet and are not included in filtered results.
Impact：Output manipulationAttack surface / entry point：Retrieval corpusAttack surface / entry point：Inference services and hardwareClear all

Another 265papers do not have category tags yet and are not included in filtered results.

  1. AttacksarXiv 2610.11932

    A Fudan team measures the citation fragility of AI search: ordinary posts can enter citations and answer text

    Measures citation fragility in AI search, demonstrating that ordinary public posts can enter citations and answers.

    arXiv
    2610.11932
    Published
    Layer
    Application layer
    Setting
    LLM applications
    Evaluated models
    Grok, Doubao, ChatGPT and others10

    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.

    Paper analysisRead analysis
  2. AttacksarXiv 2610.05894

    研究者提出闭环偏见注入攻击

    提出闭环激活引导，在扩散语言模型去噪过程中注入定向偏见

    arXiv
    2610.05894
    Published
    Layer
    Model layer
    Setting
    Models and APIs
    Attacker
    Gray-box
    Evaluated models
    LLaDA-8B-Instruct, Dream-v0-Instruct-7B, LLaDA-MoE-7B-A1B-Instruct

    Abstract通过去噪中间概率反馈动态调节残差流转向，证明扩散语言模型在推理服务栈中存在闭环偏见注入脆弱性。

    Paper analysisRead analysis
  3. DefensesarXiv 2610.05826

    研究者提出临时可见性协议，限制持续摄入 RAG 向量库的投毒暴露

    提出临时可见性协议，限制持续摄入 RAG 向量库的投毒暴露

    arXiv
    2610.05826
    Published
    Layer
    Application layer
    Setting
    LLM applications
    Abstract用截止期限制未核验可见时长。

    这篇工作给持续摄入的 RAG 向量库一个失败关闭的临时可见性协议，把协议能限制的未核验暴露和依赖检测器的中毒暴露分开。

    Paper analysisRead analysis
  5. DefensesarXiv 2610.02349

    MIRROR：面向多智能体通信的法定人数完整性防御

    提出 MIRROR，用多路径摘要多数表决保障多智能体通信完整性

    arXiv
    2610.02349
    Published
    Layer
    Application layer
    Setting
    Multi-agent systems
    Evaluated models
    Gemma-4-31B-it, Gemini 2.5 Pro
    AbstractMIRROR 靠多数路径。

    MIRROR 是 LLM 多智能体通信层上的多路径法定人数完整性原语，只处理不透明字节，不解释消息内容。要处理的是 Agent-in-the-Middle：中介合法终结 TLS 后可改在途消息，而语义校验有推理开销和误拦，签名又要求私钥管理。做法是生成结束后复制同一规范化载荷。

    Paper analysisRead analysis
  6. Evaluation and benchmarksarXiv 2609.39146

    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
    Layer
    Application layer
    Setting
    Multi-agent systems
    Evaluated models
    gpt-4o, GPT-5, GPT-6 Astra
    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
  7. InterpretabilityarXiv 2609.37737

    研究用因果激活修补定位模型服从提示注入指令的决策层

    用因果激活修补定位模型服从提示注入的决策层

    arXiv
    2609.37737
    Published
    Layer
    Model layer
    Setting
    Models and APIs
    Evaluated models
    Qwen3-4B, Qwen3-14B, Qwen3-32B and others7
    Abstract合规的因果杠杆集中在网络后三分之一的低秩子空间，该层也是混淆下检测最好的位置。

    这篇工作用逐层激活修补定位提示注入合规发生在网络何处，并把该位置用于检测验证。。

    Paper analysisRead analysis
  11. AttacksarXiv 2609.02774

    CodePoisonRAG：针对检索增强代码生成的知识投毒攻击

    提出 CodePoisonRAG，对检索增强代码生成的知识库投毒以植入漏洞

    arXiv
    2609.02774
    Published
    Layer
    Application layer
    Setting
    LLM applications
    Evaluated models
    Qwen 3.5 9B, Code Llama 13B, DeepSeek-Coder-V2 16B

    AbstractCodePoisonRAG 展示了 RACG 外部知识库的新攻击面，在较低投毒率下可定向诱发指定漏洞并部分抵御防御。

    Paper analysisRead analysis
  12. Evaluation and benchmarksarXiv 2609.02316

    Counter-GEO-Bench：现有护栏难挡 GEO 虚假信息

    构建 COUNTER-GEO-BENCH，评测生成式搜索中的 GEO 虚假信息防御

    arXiv
    2609.02316
    Published
    Layer
    Application layer
    Setting
    LLM applications
    Evaluated models
    Gemma-4-31B-IT, Qwen-3.5-35B-A3B, Llama-4-Scout-17B-16E and others8

    Abstract论文构建了针对 GEO 虚假信息的首个防御基准，揭示通用护栏的失效与反作用，并提出了有效的轻量对比检测基线。

    Paper analysisRead analysis
  14. AttacksarXiv 2609.01325

    VerTox：用可验证奖励引导的强化学习对神经排序模型实施语料投毒

    提出 VerTox，用可验证奖励强化学习生成误导文档并投毒检索语料

    arXiv
    2609.01325
    Published
    Layer
    Application layer
    Setting
    LLM applications
    Evaluated models
    SimLM-base, BGE-base, Cohere-embed-english-v3.0 and others8
    AbstractVerTox 以可验证奖励微调小 LLM，生成能抬高排序并腐蚀事实的投毒文档。

    VerTox 用可验证奖励把小 LLM 微调成语料投毒生成器，用来探测神经排序器在注入文档下的暴露面。

    Paper analysisRead analysis
  16. AttacksarXiv 2607.14493

    LogInject 框架披露 LLM 日志分析中的被动提示注入

    提出 LogInject 框架，评测日志分析中的被动提示注入及缓解

    arXiv
    2607.14493
    Published
    Layer
    Application layer
    Setting
    LLM applications
    Evaluated models
    GPT-4o, Claude 3.5 Sonnet, Llama-3-70B-Instruct

    Abstract论文评估了 LLM 日志分析中的上下文污染风险，提出分片拼接等攻击与三层防御，指出必须结合架构隔离与人工介入。

    Paper analysisRead analysis
  18. AttacksarXiv 2607.00422

    KidnapRAG：黑盒投毒攻击劫持 Agentic RAG 的推理链

    提出 KidnapRAG，用投毒文档黑盒劫持 Agentic RAG 的推理链

    arXiv
    2607.00422
    Published
    Layer
    Application layer
    Setting
    AI Agent
    Evaluated models
    Qwen2.5-32B-Instruct, Llama-3.3-70B-Instruct, QwQ-32B and others5
    Abstract顺序投毒可把多步推理拖向指定应答，作者称关键在持续引导而非单次注入。

    KidnapRAG 是针对 Agentic RAG 的 black-box 语料投毒：攻击者不能访问模型、提示词、检索器或私有轨迹，只能发布可能被检索的文档，并利用公开界面上的推理与搜索查询。

    Paper analysisRead analysis
  19. AttacksarXiv 2606.11265

    论文提出 CRCP 投毒框架，揭示 RAG 分块与重排序下的投毒失效问题

    提出 CRCP 框架，优化分块与重排下的 RAG 语料投毒

    arXiv
    2606.11265
    Published
    Layer
    Application layer
    Setting
    LLM applications
    Evaluated models
    GPT-4o, DeepSeek-R1, Qwen3-Max and others11
    AbstractCRCP 把投毒做成多阶段一致性问题。

    CRCP 研究的是带分块和重排的 RAG 语料投毒，而不是只提高文档级检索分数。

    Paper analysisRead analysis
  20. AttacksarXiv 2606.06244

    研究者提出 Ghostwriter 攻击：用伪造证据注入操纵 LLM 观点

    提出 Ghostwriter 攻击，用伪权威证据条件注入操纵模型观点

    arXiv
    2606.06244
    Published
    Layer
    Prompt layer
    Setting
    Models and APIs
    Evaluated models
    GPT-4o, Claude-3.7-Sonnet, DeepSeek-V3 and others10
    Abstract作者称攻击未被消除。

    Ghostwriter 是一种针对上下文窗口的观点引导攻击：把误导陈述改写成伪权威文本，再在相关查询中让目标模型把该观点写进回答。

    Paper analysisRead analysis
Load more