Skip to content

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

Oct 12 · Monday

All papers

Found 22 papers
Filter4

Attack type

Jailbreaks4Prompt injection4Detection evasion7

Attack technique

Gradient and representation optimization22Automated search13Multi-turn escalation5Encoding and obfuscation13Cross-modal payloads43Defense-targeted5Dormant triggers5

Impact

Harmful content3Output manipulation19

System type

Models and APIs22LLM applications3AI Agent9web agent3Embodied systems and robotics6

Attacker access

Black-box3White-box4

Attack surface / entry point

User input19Documents and messages0Images, audio and video22Weights and gradients10

Layer

Prompt layer20

System component

Vision16Audio3Multimodal20
Select any value within a category; different categories must all match. Counts include the other filters you have selected.Another 255papers have no tags yet and are not included in filtered results.
Attack technique：Gradient and representation optimizationSystem type：Models and APIsAttack surface / entry point：Documents and messagesAttack surface / entry point：Images, audio and videoClear all

Another 255papers do not have category tags yet and are not included in filtered results.

  1. 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 2610.03166

    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. Evaluation and benchmarksarXiv 2607.19855

    研究者提出最小范数攻击集成，用鲁棒性曲线替代单一 ε 评测

    提出最小范数攻击集成，用鲁棒性曲线评测图像分类器

    arXiv
    2607.19855
    Published
    Layer
    Prompt layer
    Setting
    Models and APIs
    Evaluated models
    C1–C20（CIFAR-10，Table 1）, I1–I10（ImageNet，Table 1）
    Abstract预算可控的最小范数集成用 AOI 逼近攻击前沿，DOI 给出不依赖单一ε的防御排序。

    这篇工作把对抗鲁棒评测从单一 ε、固定集成，改成可按查询预算收紧的曲线估计。

    Paper analysisRead analysis
  7. AttacksarXiv 2605.21541

    研究者提出 FRA-Attack 频域正则化方法，提升对闭源 MLLM 的迁移攻击效果

    提出 FRA-Attack，用频域对齐与梯度正则做可迁移定向视觉攻击

    arXiv
    2605.21541
    Published
    Layer
    Prompt layer
    Setting
    Models and APIs
    Attacker
    Black-box
    Evaluated models
    GPT-5.2, GPT-5.4, Claude-Opus-4.6 and others15
    AbstractFRA-Attack 以频域对齐和梯度低通做可迁移定向攻击，闭源面板上作者称整体更强，Gemma 除外。

    FRA-Attack 是面向闭源 MLLM 的迁移式定向图像攻击。作者要解决的是：空间域局部对齐重复全局低频，代理梯度的高频成分又把更新拉向代理自身。

    Paper analysisRead analysis
  10. AttacksarXiv 2605.07399

    研究者提出 GPO-V，用全局概率优化越狱扩散视觉语言模型

    提出 GPO-V，用全局概率优化的视觉扰动越狱扩散视觉语言模型

    arXiv
    2605.07399
    Published
    Layer
    Prompt layer
    Setting
    Models and APIs
    Attacker
    White-box, Black-box
    Evaluated models
    LLaDA-V, LaViDA
    Abstract作者用视觉扰动改 dVLM 的全局生成概率，以绕过两类拒绝。

    GPO-V 把越狱从“开头必须顺从”改成“整段去噪概率朝肯定方向走”。。

    Paper analysisRead analysis
  11. AttacksarXiv 2604.14604

    The AudioHijack framework achieves covert audio prompt injection against 13 speech large models

    提出 AudioHijack，用不可感知对抗音频对语音模型做上下文无关提示注入

    arXiv
    2604.14604
    Published
    Layer
    Prompt layer
    Setting
    Models and APIs
    Evaluated models
    SpeechGPT, GLM-4-Voice, VITA-Audio and others15

    Abstract提出了 AudioHijack 框架，揭示了 LALM 对听觉提示注入的脆弱性与传统防御的局限。

    Paper analysisRead analysis
  12. AttacksarXiv 2604.03117

    Unveiling Physical-World Semantic Vulnerabilities: Universal Adversarial Patches for Infrared Vision-Language Models

    Proposes the universal curved mesh patch UCGP, causing cross-task semantic bias in infrared vision-language models.

    arXiv
    2604.03117
    Published
    Layer
    Prompt layer
    Setting
    Models and APIs
    Attacker
    Gray-box
    Evaluated models
    OpenAI CLIP ViT-L/14, OpenCLIP ViT-B/16, Meta-CLIP ViT-L/14 and others10
    AbstractUCGP uses one shared curved-grid patch to simultaneously lower the classification and open-ended outputs of IR-VLM.

    UCGP is a universal physical adversarial patch for infrared vision-language models. The question is: can a patch that can be attached and reused across samples cause stable semantic deviations in the infrared visual evidence shared by classification, description, and VQA. The patch is constrained to a low-frequency curved mesh.

    Paper analysisRead analysis
  16. AttacksarXiv 2602.09431

    Grounding-driven attack: Improving the transferability of encoder-based adversarial attacks to large vision-language models

    Propose GDA, which aligns image adversarial perturbations with text-grounded evidence and transfers attacks to vision-language models

    arXiv
    2602.09431
    Published
    Layer
    Prompt layer
    Setting
    Models and APIs
    Attacker
    No-box (offline transfer)
    Evaluated models
    LLaVA, Qwen2.5-VL, InternVL3 and others10
    AbstractGDA aligns perturbations with text-grounded evidence, and the authors say this improves zero-query black-box transferability on heterogeneous LVLMs.

    GDA is a zero-query black-box encoder transfer attack for heterogeneous LVLMs.

    Paper analysisRead analysis
  17. AttacksarXiv 2602.01574

    SGHA-Attack: Semantically Guided Hierarchical Alignment for Transferable Targeted Attacks on Vision-Language Models

    提出 SGHA-Attack，对视觉语言模型做可迁移定向图像扰动

    arXiv
    2602.01574
    Published
    Layer
    Prompt layer
    Setting
    Models and APIs
    Evaluated models
    UniDiffuser, BLIP-2, InstructBLIP and others9
    Abstract用多锚点与分层对齐做定向迁移。

    SGHA-Attack 是对视觉语言模型的可迁移定向攻击：只在 CLIP 代理上优化扰动，让黑盒模型按指定文本描述图像。要处理的困难是，单一参考加上只对齐代理末层，会使优化绑在该嵌入空间上，中间层语义用不上，异构 VLM 上的指定描述更容易失败。引言把绕过安全护栏列为动机，测的则是描述是否被导向目标文本。

    Paper analysisRead analysis
  19. AttacksarXiv 2601.21531

    Adversarial robustness of large vision-language models under visual token compression

    提出 CAGE 攻击，在视觉 token 压缩下用图像扰动使预测偏离

    arXiv
    2601.21531
    Published
    Layer
    Prompt layer
    Setting
    Models and APIs
    Attacker
    Gray-box
    Evaluated models
    LLaVA, Qwen2.5-VL, InternVL3-8B and others5
    AbstractCAGE 对齐压缩瓶颈后，压缩 LVLM 的稳健准确率低于全 token 编码器攻击。

    CAGE 研究的是：LVLM 在推理时剪枝或合并视觉 token 之后，对抗稳健性该怎么估。

    Paper analysisRead analysis
Load more