Skip to content

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

Oct 12 · Monday

All papers

Found 76 papers
Filter4

Attack type

Jailbreaks23Prompt injection4Poisoning and backdoors9Model tampering9Detection evasion20

Attack technique

Gradient and representation optimization53Automated search56Multi-turn escalation22Encoding and obfuscation45Role-play and persuasion25Conversation structure manipulation11Reasoning-chain manipulation10Cross-modal payloads48Side channels3Defense-targeted24Dormant triggers23

Impact

Harmful content30Output manipulation34Availability and reliability5Oversight failure4

System type

Models and APIs69LLM applications8AI Agent34coding agent7web agent4Embodied systems and robotics7

Attacker access

Black-box16Gray-box6White-box23No-box (offline transfer)3

Attack surface / entry point

User input29Images, audio and video32Training data10Weights and gradients16Inference services and hardware4

Layer

Prompt layer43Model layer18Training and data layer10Infrastructure layer3

System component

Guardrails and judges10Monitors3Reasoning chains3Fine-tuning and open weights9Vision21Audio4Image generation3Multimodal30
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 technique：Gradient and representation optimizationAttack technique：Defense-targetedSystem type：Models and APIsSystem type：Embodied systems and roboticsClear all

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

  2. 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
  4. 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
  6. 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
  7. 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
  8. 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
  10. AttacksarXiv 2610.03430

    JIL：通过对抗后缀操纵 LLM 请求调度优先级

    提出 JIL 对抗后缀，压低长度预测以抬高请求调度优先级

    arXiv
    2610.03430
    Published
    Layer
    Infrastructure layer
    Setting
    Models and APIs
    Evaluated models
    Llama-3-8B-Instruct, Qwen2.5-7B-Instruct, Mistral-7B-Instruct-v0.3 and others4
    AbstractJIL 用对抗后缀压低长度预测，在预测式 LLM 调度中提高自身优先级。

    JIL 是对预测式 LLM 调度的攻击：用对抗后缀压低长度预测，使请求在 SRPT 式队列里更早得到服务。

    Paper analysisRead analysis
  11. 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
  12. 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
  13. AttacksarXiv 2610.01367

    研究：高质量数据筛选挡不住投毒

    提出 Bi-QSTO，优化能穿过质量筛选并在微调中削弱安全对齐的投毒样本

    arXiv
    2610.01367
    Published
    Layer
    Training and data layer
    Setting
    Models and APIs
    Evaluated models
    Llama2-7B-Chat, Llama3-8B-Instruct, Qwen2-7B-Instruct and others5
    Abstract质量筛选挡不住部分高分样本的安全削弱。

    Bi-QSTO 研究一个实际训练缺口：基于质量的数据选择会先于微调决定样本去留，而以往微调投毒多假设毒样本直接进入训练集。已对齐 LLM 的拒答会被少量有害或看似无害的微调样本削弱。作者用三个问题串起实验：选择能否排除毒样本。

    Paper analysisRead analysis
  14. AttacksarXiv 2610.01062

    Kernelized Activation Steering：将激活引导推广到再生核希尔伯特空间的新框架

    提出核化激活引导 KAS，在推理时自适应转向激活以诱导越狱

    arXiv
    2610.01062
    Published
    Layer
    Model layer
    Setting
    Models and APIs
    Evaluated models
    Qwen2.5-7B-Instruct, Gemma-2-9B-IT, Llama-3.1-8B-Instruct and others4
    AbstractKAS 用核梯度做局部自适应转向，并在线性核下恢复 DiM。

    Kernelized Activation Steering（KAS） 把推理时的激活转向写成核空间中的优化，使更新随当前激活相对源、目标参考集的位置而变化。

    Paper analysisRead analysis
  15. 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
  17. InterpretabilityarXiv 2609.24801

    研究用 XAI 归因分析 Prompt Guard 2，同义词替换可翻转其提示注入判定

    用词级归因指导同义词替换，翻转提示注入检测并验证越狱

    arXiv
    2609.24801
    Published
    Layer
    Prompt layer
    Setting
    Models and APIs
    Evaluated models
    Prompt Guard 2 (86M), Llama 3.1 8B

    Abstract作者称检测靠分散的词法线索。

    Paper analysisRead analysis
  18. AttacksarXiv 2609.15989

    Researchers propose plan injection attack that can bypass chain-of-thought monitoring

    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
  20. 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
Load more