Skip to content

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

Oct 13 · Tuesday

All papers

Found 4 papers
Filter5

Impact

Harmful content4Unauthorized actions40Data and privacy leaks19Output manipulation31Availability and reliability5

System type

AI Agent4

Attack surface / entry point

User input3

Layer

Prompt layer32Model layer6Application layer4Training and data layer4

System component

Memory2Guardrails and judges9Reasoning chains3Vision2Embeddings0Multimodal3
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.
Impact：Harmful contentLayer：Application layerSystem component：MemorySystem component：VisionSystem component：EmbeddingsClear all

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

  1. Evaluation and benchmarksarXiv 2610.03938

    Paper finds multimodal models' refusal failure rate rises by up to 68.7% after tool use is enabled

    评测 ReAct 工具调用对多模态模型拒答有害请求的削弱

    arXiv
    2610.03938
    Published
    Layer
    Application layer
    Setting
    AI Agent
    Evaluated models
    Gemini-2.5-Pro, Gemini-3.1-Pro-Preview, Claude Opus 4.6 and others13
    Abstract论文揭示了工具调用范式削弱 MLLM 拒答能力的现象，分析了稀释与偏移机制并验证了重注缓解方案。

    论文系统评估了智能体工具调用范式对多模态大语言模型拒答能力的影响，指出了工具增强视觉推理背后的安全隐患。核心探讨的问题在于：当 MLLM 从单次推理转向通过 ReAct 循环调用外部工具（打标、缩放、OCR、代码执行）时，工具使用机制本身是否会降低模型拒答有害请求的概率。

    Paper analysisRead analysis
  2. OtherarXiv 2609.12413

    SoK 综述：智能体 AI 时代的越狱攻击、防御与实践考量

    系统梳理智能体执行链路中的越狱攻击与防御并比较推理时防御

    arXiv
    2609.12413
    Published
    Layer
    Application layer
    Setting
    AI Agent
    Evaluated models
    Qwen3.5-4B, Llama-3.1-8B-Instruct
    AbstractSoK 将越狱重放到智能体流水线，并用两种模型检验回复层防御的边际与代价。

    Mia 等人把越狱安全从单轮回复扩展到智能体的规划、记忆、工具和智能体间通信，并问既有结论在更强原生对齐下是否还成立。

    Paper analysisRead analysis
You have reached the end