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

Jailbreaks4Prompt injection17Poisoning and backdoors6Extraction and theft6

Defense type

Inference-time intervention3

Impact

Harmful content4

System type

Models and APIs3

Attack surface / entry point

User input4Images, audio and video4

Layer

Prompt layer32Model layer3Application layer1

System component

Memory3Guardrails and judges12Monitors11Reasoning chains6Permissions and sandboxing5Vision4Audio3Embeddings0Multimodal7
Select any value within a category; different categories must all match. Counts include the other filters you have selected.Another 216papers have no tags yet and are not included in filtered results.
Attack type：JailbreaksLayer：Model layerLayer：Application layerSystem component：VisionSystem component：EmbeddingsClear all

Another 216papers do not have category tags yet and are not included in filtered results.

  1. DefensesarXiv 2610.09703

    研究揭示 Token-Pruning 对 VLM 安全性的影响并提出 SAP 缓解机制

    提出 Safety-Aware Pruning，缓解视觉 token 剪枝放大的多模态越狱

    arXiv
    2610.09703
    Published
    Layer
    Model layer
    Setting
    Models and APIs
    Attacker
    Black-box
    Evaluated models
    LLaVA-1.5-7B
    Abstract剪枝去掉背景 token 后注意力塌向恶意前景。

    这篇工作评估视觉 token 剪枝如何改变 VLM 在多模态越狱下的行为，并给出推理时防御 SAP。

    Paper analysisRead analysis
  3. DefensesarXiv 2602.11096

    Safety Recovery of Reasoning Models Only Needs a Few Early Steps of Guidance

    Proposes SafeThink, an inference-time defense that monitors the chain of thought and injects a redirect prefix when it falls below a threshold.

    arXiv
    2602.11096
    Published
    Layer
    Model layer
    Setting
    Models and APIs
    Evaluated models
    R1-Onevision-7B, OpenVLThinker-7B, VLAA-Thinker-7B and others6
    Paper analysisRead analysis
You have reached the end