Skip to content

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

Oct 12 · Monday

All papers

Found 3 papers
Filter2

Model risks

Deception and scheming7Misalignment and value divergence3Dangerous capabilities0Inappropriate refusal19Bias and discrimination3

System type

Models and APIs3

Layer

Model layer3
Select any value within a category; different categories must all match. Counts include the other filters you have selected.Another 265papers have no tags yet and are not included in filtered results.
Model risks：Misalignment and value divergenceModel risks：Dangerous capabilitiesClear all

Another 265papers do not have category tags yet and are not included in filtered results.

  1. InterpretabilityarXiv 2609.37616

    Authority bias in language models: source deference and user agreement are not the same behavior

    因果解耦来源依从与用户迎合的内部机制

    arXiv
    2609.37616
    Published
    Layer
    Model layer
    Setting
    Models and APIs
    Evaluated models
    Qwen3.5-27B, GPT-OSS-20B, OLMo-3.1-32B and others8

    Abstract论文发现模型对验证源的依从强于用户迎合，通过因果干预区分了二者机制，并检验了权威方向移除的缓解效果。

    Paper analysisRead analysis
  2. InterpretabilityarXiv 2609.35544

    减少谄媚能否增强拒答？机制可解释性带来的 AI 安全启示

    用 SAE 与 CFI 压低谄媚，检验直接拒答与施压下拒答是否增强

    arXiv
    2609.35544
    Published
    Layer
    Model layer
    Setting
    Models and APIs
    Evaluated models
    Qwen3.5-2B-Base, Qwen3.5-9B-Base, Qwen3.5-35B-A3B-Base
    AbstractCFI 能持久降低谄媚，但不保证直接拒答变强。

    用 SAE 选出的谄媚特征做补偿性训练注入，检验“少谄媚”是否带来“更强拒答”。。

    Paper analysisRead analysis
You have reached the end