Skip to content

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

Oct 12 · Monday

All papers

Found 8 papers
Filter4

Model risks

Deception and scheming11Reward hacking9Misalignment and value divergence8Dangerous agent actions11Dangerous capabilities18Inappropriate refusal22Hallucination and fabrication18Bias and discrimination7

Impact

Output manipulation4

System type

Models and APIs6AI Agent6coding agent2science agent1

Layer

Model layer4
Select any value within a category; different categories must all match. Counts include the other filters you have selected.Another 233papers have no tags yet and are not included in filtered results.
Model risks：Misalignment and value divergenceSystem type：Models and APIsSystem type：coding agentSystem type：science agentClear all

Another 233papers do not have category tags yet and are not included in filtered results.

  1. Evaluation and benchmarksarXiv 2610.11305

    BeliefScope：区分证据与用户压力对模型信念影响的受控框架

    用 BeliefScope 分离固定命题上证据与压力对模型信念的影响

    arXiv
    2610.11305
    Published
    Layer
    Model layer
    Setting
    Models and APIs
    Evaluated models
    Qwen3-8B, Llama3.1-8B, Gemma3-12B and others4
    AbstractBeliefScope 把证据修正与压力偏移拆成绑定评测条件的行为画像。

    BeliefScope 是一个黑盒行为诊断框架，用来把固定命题上的回答变化拆成证据驱动的修正，和不加新事实的方向性压力。信念修正评测能看到信息基础变化后的更新，谄媚评测能看到用户影响下的顺从或漂移。作者认为，当两种输入都偏向同一结论时，表面偏移本身不能区分来源。

    Paper analysisRead analysis
  2. Evaluation and benchmarksarXiv 2610.08540

    The paper proposes a scaling law framework for alignment measured by risk category.

    提出分风险对齐标度律框架，测量各类风险的对齐负担如何随规模变化

    arXiv
    2610.08540
    Published
    Layer
    Training and data layer
    Setting
    Models and APIs
    Evaluated models
    Pythia (14m–2.8b), Qwen2.5 (0.5B–72B), Qwen2.5-Instruct (0.5B–14B) and others4
    Abstract论文提出了分风险对齐标度律框架，测得真实性与倾向纠错呈标度有益，但盲后门仍存活，揭示隐藏风险的长期挑战。

    本文提出了一个将大语言模型对齐难度形式化为可测量标度关系的理论框架与预注册评测协议。

    Paper analysisRead analysis
  3. Evaluation and benchmarksarXiv 2610.06522

    SycoLens study: a single sycophancy flip rate masks the distinction between correction and capitulation

    构建 SycoLens 重放评测，区分谄媚翻转中的纠错与屈从

    arXiv
    2610.06522
    Published
    Layer
    Model layer
    Setting
    Models and APIs
    Evaluated models
    opus-5, sonnet-5, sonnet-4.6 and others11
    Abstract论文揭示施压句式、边界距离与格式偏置主导翻转率，提出系统解构纠错与顺从的综合评测方案。

    本文系统解构了大语言模型在用户质疑下的阿谀奉承行为与评测机制。

    Paper analysisRead analysis
  4. Evaluation and benchmarksarXiv 2609.39863

    FIGS：在不惩罚共情的前提下评测多轮谄媚

    提出 FIGS 双轴基准，评测多轮对话中的谄媚与校准性共情

    arXiv
    2609.39863
    Published
    Layer
    Model layer
    Setting
    Models and APIs
    Evaluated models
    DeepSeek V4.1 Flash, Gemini 3.8 Flash, GLM 5.3 and others8
    AbstractFIGS 把守事实和恰当共情分开计分，事实向提示会换来更冷的失败。

    FIGS 是一个固定的多轮评测套件，用来同时看模型会不会在压力下放弃事实，以及会不会在守住事实时忽略用户实际说过的感受。

    Paper analysisRead analysis
  5. Evaluation and benchmarksarXiv 2609.32616

    Cave-Bench：虚假指控下 LLM 智能体最高 60.06% 的运行会破坏正确工作

    提出 CAVE-BENCH，评测智能体在虚假指责下破坏已完成的正确工作

    arXiv
    2609.32616
    Published
    Layer
    Application layer
    Setting
    coding agent
    Evaluated models
    Claude-Sonnet-5, Claude-Opus-5, GPT-5.6-Sol and others14

    Abstract论文提出了评估智能体在虚假指责下破坏已有正确成果的 CAVE-BENCH，发现强模型易推翻已知证据，提出了门控缓解方案。

    Paper analysisRead analysis
  6. Evaluation and benchmarksarXiv 2609.29429

    Jev 被用于零样本检测十类对齐失败，中位 AUROC 达 0.886

    提出 RLCDAlignBench，零样本检测多种对齐失败

    arXiv
    2609.29429
    Published
    Layer
    Prompt layer
    Setting
    Models and APIs
    Evaluated models
    Jev (jev-1.13.0)
    Abstract一次通用问题已能给多数对齐失败排序，大的误差来自状态和标签。

    RLCDAlignBench 测量用强化学习做校准决策的 Jev，能否当作对齐失败的零样本检测器。

    Paper analysisRead analysis
You have reached the end