Skip to content

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

Oct 12 · Monday

All papers

Found 65 papers
Filter3

Attack type

Jailbreaks34Prompt injection4Poisoning and backdoors6Detection evasion14

Attack technique

Gradient and representation optimization45Automated search52Multi-turn escalation22Encoding and obfuscation39Role-play and persuasion23Conversation structure manipulation11Reasoning-chain manipulation9Cross-modal payloads43Side channels3Defense-targeted23Dormant triggers17

Impact

Harmful content36Output manipulation21Oversight failure4

System type

Models and APIs65LLM applications4AI Agent31coding agent8web agent4Embodied systems and robotics4

Attacker access

Black-box30Gray-box5White-box6

Attack surface / entry point

User input34Images, audio and video47Training data7Weights and gradients6

Layer

Prompt layer49Model layer7Training and data layer8

System component

Guardrails and judges14Monitors3Reasoning chains4Fine-tuning and open weights7Vision34Audio6Image generation6Multimodal46
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.
Attack technique：Cross-modal payloadsAttack technique：Defense-targetedSystem type：Models and APIsClear all

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

  2. 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
  3. AttacksarXiv 2610.06670

    研究者提出 Reward Stealing Attack，从对齐模型行为反推安全奖励实施攻击

    提出 ReSA，从对齐模型行为恢复安全奖励并在解码时反转

    arXiv
    2610.06670
    Published
    Layer
    Model layer
    Setting
    Models and APIs
    Attacker
    White-box, Gray-box
    Evaluated models
    Llama3.1-8B-Instruct, Gemma-7B-Instruct, Qwen2.5-7B-Instruct and others9
    AbstractReSA 用最大熵 IRL 从对齐模型行为恢复代理安全奖励，并在解码时反转以诱导有害输出。

    ReSA 是一种对抗攻击：从对齐模型的可观察行为恢复代理安全奖励，再在解码时把该奖励反过来用。。

    Paper analysisRead analysis
  5. AttacksarXiv 2610.03166

    LiBRA：通过双向隐空间优化实现检测感知的图像水印去除

    提出 LiBRA，在潜空间双向优化以规避图像水印双侧检测

    arXiv
    2610.03166
    Published
    Layer
    Model layer
    Setting
    Models and APIs
    Attacker
    White-box
    Evaluated models
    Stable Signature, YU1, HiDDeN and others4

    AbstractLiBRA 用双向潜空间优化规避双侧水印检测，并提高相对反转基线的保真。

    Paper analysisRead analysis
  6. AttacksarXiv 2610.03124

    研究提出 Untag 攻击框架，使开源大模型触发标签滥用检测机制失效

    提出 UNTAG，绕过开源权重模型的 trigger-tag 滥用检测

    arXiv
    2610.03124
    Published
    Layer
    Model layer
    Setting
    Models and APIs
    Attacker
    Black-box, White-box
    Evaluated models
    Llama2 (7B), Llama3 (8B), Qwen2.5 (7B)
    Abstract作者用 UNTAG 评估 trigger-tag，认为其不能充当开放权重滥用检测的主要边界。

    UNTAG 是针对开放权重 LLM 上条件滥用检测 trigger-tag 的统一攻击评估：信号可以在解码期注入，也可以写进权重，但攻击者若能改输出或改已发布权重，检测可以失效。

    Paper analysisRead analysis
  8. AttacksarXiv 2609.27124

    Fed-ADR：用恶意编排服务器协同客户端梯度操纵攻击联邦学习

    提出 Fed-ADR，用编排服务器协同操纵梯度攻破联邦学习并检测恢复

    arXiv
    2609.27124
    Published
    Layer
    Training and data layer
    Setting
    Models and APIs
    Attacker
    Gray-box
    Evaluated models
    CNN（MNIST：2 Conv + 3 FC）, CNN（Fashion-MNIST：6 Conv + 1 FC）, ResNet18（CIFAR-10）
    AbstractFed-ADR 用 gray-box 编排攻击绕过多种聚合防御，再用 Hessian 一致性检测和窗口内恢复把准确率拉回。

    Fed-ADR 是一套面向联邦学习的攻击、检测与恢复框架：编排服务器改写合谋客户端的梯度，使参数服务器的现有防御把投毒更新当成良性，同时在发现后用历史梯度把全局模型原地修回来。

    Paper analysisRead analysis
  10. AttacksarXiv 2609.15989

    研究者提出 plan injection 攻击，可绕过思维链监控

    提出 plan injection 攻击，植入良性伪装计划以绕过思维链监视器

    arXiv
    2609.15989
    Published
    Layer
    Prompt layer
    Setting
    Models and APIs
    Attacker
    Gray-box
    Evaluated models
    Qwen2.5-Coder-7B-Instruct, Qwen3-8B, DeepSeek-R1-0528 and others6

    Abstract论文指出模型会将外部上下文中的伪装计划改写为自身推理，使思维链监视器失效甚至反向合理化恶意行为。

    Paper analysisRead analysis
  12. AttacksarXiv 2609.10117

    研究揭示基于混淆的端侧 LLM 保护方案的安全边界

    提出 Collapse，从端侧混淆暴露视图抽取高效用替代模型

    arXiv
    2609.10117
    Published
    Layer
    Model layer
    Setting
    Models and APIs
    Evaluated models
    BERT-Base, ViT-Base, Qwen2.5-0.5B and others4
    AbstractOprior 是四原语复合的安全边界，Collapse 利用列方向泄漏抽取替代模型，Oext 只被同一攻击评测。

    这篇工作用代数原语统一若干 TEE 卸载混淆，给出该族的典范边界，再说明该边界挡不住一种分阶段模型抽取。。

    Paper analysisRead analysis
  13. AttacksarXiv 2609.08236

    研究：内容不变的风格包装可翻转 LLM 安全评判器的判定

    提出内容不变样式包装，翻转安全评审器的判定

    arXiv
    2609.08236
    Published
    Layer
    Prompt layer
    Setting
    Models and APIs
    Attacker
    Black-box
    Evaluated models
    GPT-4o-mini (gpt4o), GPT-4o-mini (strongreject), DeepSeek-Chat and others7
    Abstract论文揭示特定安全评审器易受保持内容不变的外层包装诱导改变判定，并指出安全评估应将判定稳定性作为重要考量指标。

    这是一项关于大语言模型自动化安全评审系统表面样式脆弱性的评估研究。

    Paper analysisRead analysis
  15. AttacksarXiv 2609.07216

    DIVA：利用视觉锚点对视频生成模型实施多模态越狱

    提出 DIVA，将有害意图拆成视觉锚点与运动文本以越狱视频生成模型

    arXiv
    2609.07216
    Published
    Layer
    Prompt layer
    Setting
    Models and APIs
    Attacker
    Black-box
    Evaluated models
    Wan, CogVideoX, LTX-Video and others9
    AbstractDIVA 用视觉锚点抑制语义漂移做视频越狱，并给出多条件安全基准 TI2VSafetyBench。

    DIVA 是针对带参考图的视频生成模型的免训练 black-box 越狱：用一致性机制本身堵住语义漂移这条安全逃逸路径。。

    Paper analysisRead analysis
Load more