Skip to content

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

Oct 12 · Monday

All papers

Found 190 papers
Filter4

Attack type

Jailbreaks58Prompt injection54Poisoning and backdoors71Model tampering8Extraction and theft3Detection evasion17

Attack technique

Gradient and representation optimization68Automated search115Multi-turn escalation30Encoding and obfuscation64Role-play and persuasion34Conversation structure manipulation16Reasoning-chain manipulation14Cross-modal payloads58Side channels4Defense-targeted44Dormant triggers58Selection manipulation12

Defense type

Detection and filtering3

Impact

Harmful content69Unauthorized actions26Data and privacy leaks16Output manipulation70Availability and reliability6Oversight failure3

System type

Models and APIs103LLM applications18AI Agent70coding agent15web agent5Multi-agent systems6Embodied systems and robotics10

Attacker access

Black-box39Gray-box3White-box21No-box (offline transfer)3

Attack surface / entry point

User input83Web pages and interfaces10Documents and messages14Tool responses11Tool and skill descriptions17Code repositories and dependencies7Retrieval corpus12Memory writes17Inter-agent messages4Images, audio and video54Training data17Weights and gradients26Distribution channels28

Layer

Prompt layer78Model layer14Application layer76Training and data layer20

System component

MCP and skills20Memory24RAG12Guardrails and judges26Reasoning chains7Permissions and sandboxing3Fine-tuning and open weights21Vision42Audio8Image generation9Embeddings4Multimodal59
Select any value within a category; different categories must all match. Counts include the other filters you have selected.Another 254papers have no tags yet and are not included in filtered results.
Attack technique：Gradient and representation optimizationAttack technique：Encoding and obfuscationAttack technique：Dormant triggersAttack technique：Selection manipulationClear all

Another 254papers do not have category tags yet and are not included in filtered results.

  1. AttacksarXiv 2610.13928

    AI models can provably hide arbitrary capabilities

    Proposes digital lock compilation, encrypting executable networks into host weights to hide arbitrary capabilities

    arXiv
    2610.13928
    Published
    Layer
    Model layer
    Setting
    Models and APIs
    Evaluated models
    Gemma 3 270M
    AbstractThe authors use a digital lock to hide an executable capability and state that elicitation on the chemistry and parity tasks did not recover the capability.

    The authors construct a class of backdoors that encrypt an executable network into the weights, used to discuss the limits of white-box elicitation audits under supply chain attacks.

    Paper analysisRead analysis
  2. AttacksarXiv 2610.12644

    Cancerous backdoor attack: progressive activation via invisible tokens and a stage controller.

    Proposes the Cancerous backdoor, which progressively corrupts reasoning using invisible stage tokens.

    arXiv
    2610.12644
    Published
    Layer
    Training and data layer
    Setting
    Models and APIs
    Evaluated models
    Llama-3.1-8B-Instruct, Qwen2.5-7B-Instruct, Mistral-7B-Instruct-v0.3
    AbstractHidden tokens plus a stage controller make the backdoor progressively corrupt reasoning upon triggering, and the authors believe that looking only at the answer will miss it.

    Cancerous Backdoor Attack changes LLM backdoors from one-time activation to multi-stage reasoning corruption that accumulates progressively with triggers.

    Paper analysisRead analysis
  3. Evaluation and benchmarksarXiv 2610.11766

    研究提出意图恢复率作为 LLM 安全评测的补充指标

    提出 IRIS，用意图恢复率区分遮蔽越狱下无害输出的原因

    arXiv
    2610.11766
    Published
    Layer
    Prompt layer
    Setting
    Models and APIs
    Evaluated models
    GPT-4o, Claude Sonnet 4.5, Claude Opus 4.5 and others6
    Abstract无害输出的证据权重取决于任务是否被恢复。

    IRIS 把意图恢复当作中间信号，用来解释意图遮蔽提示下为何没有出现有害协助。

    Paper analysisRead analysis
  4. AttacksarXiv 2610.12524

    An exploration of attacks on AI agent systems

    重测并扩展对 AI Agent 的提示注入，并用 Agent Card 欺骗操纵 A2A 选择

    arXiv
    2610.12524
    Published
    Layer
    Application layer
    Setting
    AI Agent
    Evaluated models
    Ollama Mistral 0.9.5, llama3-8b-8192, gemini-2.0-flash-lite and others5
    Abstract单智能体过滤与分类可被格式或措辞绕过，A2A 选卡还会偏向名称和描述。

    作者在自建的 LangChain、AutoGen 和 A2A 系统上重测提示注入，并改 Agent Card 看选择是否偏移。摘要称试验主要凸显单智能体弱点，同时把攻击扩到多智能体。

    Paper analysisRead analysis
  6. AttacksarXiv 2610.09819

    FAB 攻击：给 HuBERT/WavLM 植入后门，背景警笛声即可让下游语音任务失灵

    提出 FAB 攻击，向声学基础模型植入可由环境声触发的后门

    arXiv
    2610.09819
    Published
    Layer
    Training and data layer
    Setting
    Models and APIs
    Evaluated models
    HuBERT-base, WavLM-base

    Abstract论文提出了针对声学基模的 FAB 后门攻击，在弱威胁模型下实现了任务无关、物理可触发的后门植入。

    Paper analysisRead analysis
  8. AttacksarXiv 2610.09240

    WebMirage hijacks visual web agents' browser operations with adversarial images

    提出 WebMirage，用局部对抗图像劫持视觉网页 Agent 的浏览器操作

    arXiv
    2610.09240
    Published
    Layer
    Application layer
    Setting
    web agent
    Evaluated models
    LLaVA-v1.5-13B, LLaVA-v1.6-34B, MiniCPM-o-8B and others6
    Abstract论文提出了针对网络智能体的端到端视觉红队框架 WebMirage，揭示了从视觉定位到浏览器执行的安全脆弱性。

    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。

    Paper analysisRead analysis
  9. Evaluation and benchmarksarXiv 2610.09033

    研究发布 ASRD 数据集，评测五款开源模型在非规范输入下的安全表现

    arXiv
    2610.09033
    Published
    Layer
    Prompt layer
    Setting
    Models and APIs
    Evaluated models
    Gemma 2 9B, Gemma 3 4B, Llama 3.1 8B and others6
    Abstract五模型、七族表面形式的四态评测里，低有害遵从不等于读懂后的安全处理。

    ASRD 用四态标签检查五个开源模型如何处理非规范表面上的有害请求。

    Paper analysisRead analysis
  10. AttacksarXiv 2610.09027

    P-VMI：对抗图像经 KV cache 持续影响多轮对话

    提出持久视觉记忆注入，使对抗图像经缓存在触发时操控输出

    arXiv
    2610.09027
    Published
    Layer
    Infrastructure layer
    Setting
    Models and APIs
    Evaluated models
    Qwen3-8B, Qwen3-VL-8B-Instruct, Gemma-4-12B

    Abstract论文提出 P-VMI，证实对抗输入被掩蔽后仍可经 KV 缓存持久控制模型，强调了保留状态的安全风险。

    Paper analysisRead analysis
  13. AttacksarXiv 2610.07723

    Researchers propose answer-side backdoors: letting the model self-generate trigger words to bypass safety refusals.

    提出回答端后门，让模型自生成触发词以在干净有害请求下绕过拒答

    arXiv
    2610.07723
    Published
    Layer
    Training and data layer
    Setting
    Models and APIs
    Evaluated models
    Llama-2-7B-chat-hf, Mistral-7B-Instruct-v0.3, Gemma-7B-it and others4

    Abstract论文提出了多轮对话中回答端自生成触发词绕过安全拒绝的后门攻击，指出输入端防护存在盲区。

    Paper analysisRead analysis
  14. AttacksarXiv 2610.07654

    研究揭示策略蒸馏的安全后门风险

    测量带后门教师经 OPD 向干净学生迁移潜伏后门

    arXiv
    2610.07654
    Published
    Layer
    Training and data layer
    Setting
    Models and APIs
    Evaluated models
    Qwen2.5-1.5B-Instruct, Qwen2.5-3B-Instruct, Qwen2.5-7B-Instruct
    Abstract作者称 OPD 可在压低无触发词有害率时迁入后门。

    带后门的外部教师经 OPD，可在压低无触发词有害率的同时，把触发条件下的有害行为传给初始干净学生。。

    Paper analysisRead analysis
  15. AttacksarXiv 2610.07645

    SkillPoison：用成功经验投毒自进化 Agent 的技能形成

    提出 SkillPoison，用局部有效的成功经验诱导自进化 Agent 形成过度泛化技能

    arXiv
    2610.07645
    Published
    Layer
    Application layer
    Setting
    AI Agent
    Evaluated models
    deepseek-v4-flash, Qwen V3.8 Flash, GPT-5.5
    AbstractSkillPoison 通过局部有效且验证成功的经验诱导技能过度泛化，在多个评测基准上实现了持久的下游攻击影响。

    SkillPoison 提出了一种针对自进化智能体经验到技能形成过程的攻击范式，通过组织已验证的成功经验诱导技能过度泛化。

    Paper analysisRead analysis
  16. AttacksarXiv 2610.07510

    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
    Layer
    Application layer
    Setting
    coding agent
    Evaluated models
    Qwen2.5-Coder-3B-Instruct, Qwen2.5-Coder-7B-Instruct, Qwen3-Coder-30B-A3B-Instruct
    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
  18. AttacksarXiv 2610.06083

    研究者提出面向深度强化学习的轨迹级可迁移黑盒对抗攻击

    提出轨迹级可迁移对抗攻击，降低深度强化学习策略的期望回报

    arXiv
    2610.06083
    Published
    Layer
    Model layer
    Setting
    Embodied systems and robotics
    Evaluated models
    DQN, Double DQN (DDQN)
    Abstract轨迹级后退视界攻击在 CartPole 上比逐步 FGSM 类和 Sign 噪声更压低回报。

    这项工作研究 DRL 上的迁移式黑盒观测攻击：受害参数不可访问时，在白盒代理上造扰动再迁移。

    Paper analysisRead analysis
  19. AttacksarXiv 2610.06049

    研究者提出仅修改 SAE 解码器的后门攻击，可在不改动语言模型的情况下植入行为

    提出只改 SAE 解码器的后门，在冻结语言模型上植入代码插入

    arXiv
    2610.06049
    Published
    Layer
    Model layer
    Setting
    Models and APIs
    Evaluated models
    Gemma 3 4B Instruct, Qwen3 1.7B, Llama 3.1 8B Instruct
    Abstract只改解码器的 SAE 可带上代码插入后门，且若干常规质量指标变化不大。

    只改 SAE 解码器、冻结编码器与语言模型，即可把代码插入行为放进推理时的辅助组件。作者把这视为 SAE 供应链问题：checkpoint 一旦替换某一层激活，就参与后续计算，而不只是解释工具。

    Paper analysisRead analysis
  20. 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
Load more