Skip to content

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

Oct 13 · Tuesday

All papers

Found 4 papers
Filter5

Attack type

Poisoning and backdoors3

System type

Models and APIs3

Attack surface / entry point

User input134Web pages and interfaces26Documents and messages28Tool responses28Tool and skill descriptions16Code repositories and dependencies7Retrieval corpus16Memory writes40Inter-agent messages7Images, audio and video89Training data12Weights and gradients4Distribution channels11

Layer

Training and data layer4

System component

Memory0Guardrails and judges2Fine-tuning and open weights37Vision2Image generation6Embeddings0Multimodal9
Select any value within a category; different categories must all match. Counts include the other filters you have selected.Another 215papers have no tags yet and are not included in filtered results.
Attack surface / entry point：Weights and gradientsSystem component：MemorySystem component：Guardrails and judgesSystem component：VisionSystem component：EmbeddingsClear all

Another 215papers do not have category tags yet and are not included in filtered results.

  1. DefensesarXiv 2610.09941

    OrthoPurify 用一步正交投影净化被后门污染的视觉语言模型

    提出 OrthoPurify，用正交投影清除视觉语言模型适配器中的劫持方向

    arXiv
    2610.09941
    Published
    Layer
    Training and data layer
    Setting
    Models and APIs
    Evaluated models
    LLaVA-1.5-7B, Qwen3-VL-8B, LLaVA-1.5-13B and others5
    Abstract一次投影去掉劫持方向。

    OrthoPurify 是部署前的 LVLM 后门净化方法，用一次正交投影去掉被劫持的权重更新方向。

    Paper analysisRead analysis
  2. DefensesarXiv 2609.15671

    QPriv-VL：面向视觉语言模型的问题引导 token 裁剪隐私防御

    提出 QPriv-VL，用问题引导剪枝降低分布式视觉问答的隐私泄露

    arXiv
    2609.15671
    Published
    Layer
    Training and data layer
    Setting
    Models and APIs
    Evaluated models
    LLaVA-1.5-7B, Llama-2
    AbstractQPriv-VL 在切层前按问题相关度与 DINOv2 敏感度剪视觉 token，作者称能在约四成 token 下兼顾准确率与四类攻击。

    QPriv-VL 把问题引导的视觉 token 剪枝做成分布式 VQA 的客户端隐私防御，而不是只做集中式推理加速。

    Paper analysisRead analysis
  3. AttacksarXiv 2604.03890

    From prompts to physical actions: structured backdoor attacks against LLM-mediated robot control systems

    Propose a structured JSON backdoor that triggers malicious actions in robot control via a poisoned LoRA

    arXiv
    2604.03890
    Published
    Layer
    Training and data layer
    Setting
    Embodied systems and robotics, LLM applications
    Evaluated models
    Llama-3.1-8B-Instruct, Gemma-2-9B-IT, DeepSeek-R1-Distill-Llama-8B and others4

    Abstract推理层后门传不到执行。

    Paper analysisRead analysis
  4. AttacksarXiv 2602.01658

    Efficient Adversarial Attacks on High-Dimensional Offline Bandits

    提出三类攻击，训练前扰动奖励模型以劫持离线 bandit

    arXiv
    2602.01658
    Published
    Layer
    Training and data layer
    Setting
    Models and APIs
    Evaluated models
    Image Reward, Aesthetic Model
    Abstract公开奖励权重可被小扰动劫持离线 bandit，打乱日志只部分缓解。

    作者研究离线 bandit 在奖励模型被事先改权重时，还会不会选出最优臂。

    Paper analysisRead analysis
You have reached the end