Skip to content

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

Oct 12 · Monday

All papers

Found 4 papers
Filter3

Attack type

Jailbreaks1Prompt injection6Model tampering3Detection evasion8Malicious use4

Impact

Harmful content232Unauthorized actions17Data and privacy leaks6Output manipulation10Oversight failure4Increased misuse capability3

System type

Models and APIs4

Attack surface / entry point

Weights and gradients3

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.
Attack type：JailbreaksAttack type：Model tamperingImpact：Oversight failureClear all

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

  1. AttacksarXiv 2610.03124

    研究提出 Untag 攻击框架，使开源大模型触发标签滥用检测机制失效

    提出 UNTAG，绕过开源权重模型的 trigger-tag 滥用检测

    arXiv
    2610.03124
    Published
    Layer
    Model layer
    Setting
    Models and APIs
    Attacker
    Black-box, White-box
    Evaluated models
    Llama2 (7B), Llama3 (8B), Qwen2.5 (7B)
    Abstract作者用 UNTAG 评估 trigger-tag，认为其不能充当开放权重滥用检测的主要边界。

    UNTAG 是针对开放权重 LLM 上条件滥用检测 trigger-tag 的统一攻击评估：信号可以在解码期注入，也可以写进权重，但攻击者若能改输出或改已发布权重，检测可以失效。

    Paper analysisRead analysis
  2. Evaluation and benchmarksarXiv 2609.08901

    论文揭示 BatchNorm 架构下机器遗忘评测的归一化伪影

    诊断 BatchNorm 架构下机器遗忘评测的归一化伪影

    arXiv
    2609.08901
    Published
    Layer
    Model layer
    Setting
    Models and APIs
    Evaluated models
    ResNet-18, ResNet-50, ViT-S/16 and others4
    Abstract一次不改权重的 BN 重校准能翻转多数方法的表面遗忘，但不改变权重里的编码器泄露。

    BN illusion 是作者对近似机器遗忘评测中一种 BatchNorm 测量伪影的命名：不改权重、只用 retain 数据重算运行统计，就可以把表面遗忘准确率翻过来。

    Paper analysisRead analysis
  3. Evaluation and benchmarksarXiv 2609.00498

    SEAV：面向大语言模型越狱评测的有效性验证框架

    提出 SEAV，逐步核验越狱回复的事实与操作有效性

    arXiv
    2609.00498
    Published
    Layer
    Prompt layer
    Setting
    Models and APIs
    AbstractSEAV 用逐步检索和依赖检查重判越狱成功，SD-A 与三个基准上的原成功标签都有一批被改判。

    SEAV 是一层生成后的越狱评测：回复要相关、事实正确、顺序合规，并且在操作上足以推进有害意图，才算成功。。

    Paper analysisRead analysis
  4. AttacksarXiv 2608.29458

    Reference-Grafting 在提取沙袋行为隐藏能力上追平微调

    提出参考嫁接，在推理阶段用激活干预诱导沙袋隐藏能力

    arXiv
    2608.29458
    Published
    Layer
    Model layer
    Setting
    Models and APIs
    Attacker
    White-box
    Evaluated models
    Qwen2.5-1.5B, Qwen2.5-3B, Qwen2.5-14B and others15

    Abstract提出参考嫁接在推理阶段以低成本匹配微调诱导效果，揭示隐瞒表征重叠与阈值门控机制。

    Paper analysisRead analysis
You have reached the end