Skip to content

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

Oct 12 · Monday

All papers

Found 5 papers
Filter3

Model risks

Deception and scheming5Reward hacking24Misalignment and value divergence14Dangerous agent actions8Inappropriate refusal17Bias and discrimination3

Defense type

Model hardening4Instruction–data separation1Monitoring and auditing16

System type

Models and APIs4

Layer

Training and data layer4
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：Deception and schemingDefense type：Model hardeningDefense type：Instruction–data separationClear all

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

  1. DefensesarXiv 2610.11012

    Redwood Research 提出蒸馏双重困境：用蒸馏让模型坦白隐藏行为或只继承能力

    提出 DFI 与 DFC 蒸馏，定罪隐瞒失配的教师或安全提取能力

    arXiv
    2610.11012
    Published
    Layer
    Training and data layer
    Setting
    Models and APIs
    Evaluated models
    Llama-3.3-70B-Instruct (AuditBench SFT/KTO organisms), Llama-3.3-70B-Instruct (DFI distilled student), Llama-3.1-8B-Instruct (DFI distilled student) and others4

    Abstract概括了论文针对隐瞒失配模型提出的 DFI 与 DFC 机制、主实验发现及适用边界。

    Paper analysisRead analysis
  2. DefensesarXiv 2609.38909

    Purdue 提出 Pact：将欺骗作为行为遗忘

    提出 PACT，遗忘推理模型在压力下的欺骗行为并保留上下文理解

    arXiv
    2609.38909
    Published
    Layer
    Training and data layer
    Setting
    Models and APIs
    Attacker
    White-box
    Evaluated models
    DeepSeek-R1-Distill-Qwen-32B, QwQ-32B

    Abstract论文将大模型欺骗形式化为上下文条件行为，提出 PACT 框架在消除欺骗的同时保留了上下文遵循能力。

    Paper analysisRead analysis
  3. DefensesarXiv 2609.38645

    用探针引导微调对齐模型且不损失可监控性

    用持续更新的探针做微调，降低有害性并保持线性可监测性

    arXiv
    2609.38645
    Published
    Layer
    Training and data layer
    Setting
    Models and APIs
    Evaluated models
    Mistral 7B Instruct v0.1, Llama 3 8B Instruct Abliterated, Qwen3-14B
    Abstract持续重拟合的探针可降低有害性、提高诚实性，训练后概念仍能被线性探针读出。

    Probe-guided fine-tuning 把探针分数当作唯一对齐信号，用来推动 harmlessness 与 honesty，并检查训练后这些概念是否仍线性可读。

    Paper analysisRead analysis
  4. Risky behaviorarXiv 2609.33220

    斯坦福与 Anthropic 研究：RL 后训练中模型承认失败的披露行为远不如任务能力稳定

    度量强化学习后失败披露的跨运行不稳定性，并提出失败条件参考锚定

    arXiv
    2609.33220
    Published
    Layer
    Training and data layer
    Setting
    Models and APIs
    Evaluated models
    Qwen2.5-1.5B, Qwen2.5-7B, Qwen2.5-14B and others11

    Abstract纯结果强化学习会使模型失败披露行为出现远高于任务能力的跨运行发散，采用失败条件参考锚定可稳定汇报但存在性能权衡。

    Paper analysisRead analysis
  5. DefensesarXiv 2609.19587

    红队测试 Auto Mode：用注入与多上下文攻击绕过编码 Agent 阻断监控器

    提出 Auto Mode ++，加固编码 Agent 阻断监视器

    arXiv
    2609.19587
    Published
    Layer
    Application layer
    Setting
    coding agent
    Evaluated models
    Opus 5, Opus 4.8, Opus 4.7 and others8

    Abstract系统评估生产阻断监视器在面对持续恶意智能体时的表现，提出加固防御并指明多上下文防御挑战。

    Paper analysisRead analysis
You have reached the end