Skip to content

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

Oct 12 · Monday

Defenses and guardrails · Papers

Found 5 papers
Filter2

Attack type

Jailbreaks4

Defense type

Model hardening3

Impact

Harmful content5

System type

Models and APIs5

Attack surface / entry point

User input4

System component

MCP and skills0Guardrails and judges19Monitors12Reasoning chains5Fine-tuning and open weights3Vision9Multimodal12
Select any value within a category; different categories must all match. Counts include the other filters you have selected.
System component：MCP and skillsSystem component：Reasoning chainsTopic: Defenses and guardrailsClear all
  2. DefensesarXiv 2609.36562

    ThinkingGuard：通过逐步风险归因解码多模态大语言模型中的隐式危害

    构建 TriggerBench 并训练 ThinkingGuard，检测图文隐式风险

    arXiv
    2609.36562
    Published
    Layer
    Prompt layer
    Setting
    Models and APIs
    Evaluated models
    ThinkingGuard（Qwen3-VL-8B-Instruct）
    AbstractTriggerBench 隔离触发条件，ThinkingGuard 用分步归因在隐式风险检测上取得作者报告的较高平均分。

    这篇工作研究多模态隐式风险检测：图文各自良性，风险只在关键元素与触发元素交互时出现。

    Paper analysisRead analysis
  4. DefensesarXiv 2604.00012

    研究提出 SafeReAct：恢复后训练大模型被掩盖的安全机制

    提出 SafeReAct，用 LoRA 重激活后训练模型被掩盖的安全机制

    arXiv
    2604.00012
    Published
    Layer
    Model layer
    Setting
    Models and APIs
    Evaluated models
    DeepSeek-R1-distilled LLaMA-8B, DeepSeek-R1-distilled Qwen-7B, DeepSeek-R1-distilled Qwen-14B and others9
    AbstractSafeReAct 用剪枝得到的安全表示做 LoRA 对齐，以恢复后训练模型被掩盖的拒绝行为。

    这篇工作针对后训练 LLM 的安全下降，提出表示对齐方法SafeReAct。

    Paper analysisRead analysis
You have reached the end