Skip to content

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

Oct 12 · Monday

Defenses and guardrails · Papers

Found 2 papers
Filter2

Defense type

Model hardening36Inference-time intervention15Detection and filtering19Instruction–data separation2Permissions and approval0Monitoring and auditing10
Select any value within a category; different categories must all match. Counts include the other filters you have selected.
Defense type：Instruction–data separationDefense type：Permissions and approvalTopic: Defenses and guardrailsClear all
  1. DefensesarXiv 2609.10613

    后验重加权框架解释并缓解多模态大模型上下文越狱

    用后验重加权解释多模态上下文越狱，并提出风险门控的良性示例注入防御

    arXiv
    2609.10613
    Published
    Layer
    Prompt layer
    Setting
    Models and APIs
    Evaluated models
    Qwen3-VL-2B, Qwen3-VL-4B, Qwen3-VL-8B and others8
    Abstract后验重加权解释上下文越狱的缩放，并支持风险门控的良性示例防御。

    Zhang 等把多模态 ICL 越狱写成对齐模型在安全与有害隐模式之间的推理时后验漂移，并据此做风险门控防御。

    Paper analysisRead analysis
  2. DefensesarXiv 2609.02786

    SafeEvolve：基于智能体经验的 Harness-Policy 协同演化安全对齐框架

    提出 SafeEvolve，用轨迹证据协同演化安全 harness 与策略

    arXiv
    2609.02786
    Published
    Layer
    Application layer
    Setting
    AI Agent
    Evaluated models
    Qwen3.5-4B, Qwen3-4B-Instruct-2507, Qwen3.5-1.7B and others4
    AbstractSafeEvolve 用轨迹证据协同更新安全 harness 和策略，在两个 4B 模型的智能体安全基准上改变安全与效用。

    SafeEvolve 把已完成的 on-policy 轨迹变成安全 harness 的有界更新，并用两阶段 SFT-RL 把这些更新内化进工具使用策略。

    Paper analysisRead analysis
You have reached the end