Skip to content

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

Oct 13 · Tuesday

All papers

Found 3 papers
Filter6

Attack type

Prompt injection10Poisoning and backdoors3Model tampering3Detection evasion3

System type

Models and APIs3

Attack surface / entry point

Documents and messages0Weights and gradients3

System component

Monitors1Reasoning chains2Fine-tuning and open weights14Embeddings0
Select any value within a category; different categories must all match. Counts include the other filters you have selected.Another 213papers have no tags yet and are not included in filtered results.
Attack type：Model tamperingAttack surface / entry point：Documents and messagesAttack surface / entry point：Weights and gradientsSystem component：MonitorsSystem component：Reasoning chainsSystem component：EmbeddingsClear all

Another 213papers do not have category tags yet and are not included in filtered results.

  1. 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 2608.05578

    AMS 工具通过激活分析检测语言模型的安全训练改动

    提出 AMS，用激活分离与方向相似度检测安全训练改动

    arXiv
    2608.05578
    Published
    Layer
    Model layer
    Setting
    Models and APIs
    Evaluated models
    Llama-3.1-8B-Instruct, Llama-3.2-3B-Instruct, Llama-3.2-1B-Instruct and others14
    AbstractAMS 按激活签名把安全改动分四类，前三类可由两级扫描覆盖，行为微调会漏。

    AMS 是面向开放权重模型、部署前使用的激活扫描器，用来看安全训练被拿掉或改写之后，激活空间里还留着什么几何痕迹。

    Paper analysisRead analysis
You have reached the end