Skip to content

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

Oct 12 · Monday

All papers

Found 5 papers
Filter3

Attack type

Jailbreaks5Prompt injection18Poisoning and backdoors54Model tampering15Extraction and theft8Detection evasion4Malicious use3

Impact

Harmful content4

System type

AI Agent3

Attack surface / entry point

User input239Tool responses15Tool and skill descriptions4Code repositories and dependencies2Images, audio and video47Training data3Weights and gradients7

Layer

Application layer3
Select any value within a category; different categories must all match. Counts include the other filters you have selected.Another 261papers have no tags yet and are not included in filtered results.
Attack type：JailbreaksAttack surface / entry point：Code repositories and dependenciesAttack surface / entry point：Training dataClear all

Another 261papers do not have category tags yet and are not included in filtered results.

  1. DefensesarXiv 2609.02293

    SEAL：通过共享专家对齐强化 MoE 全局安全

    提出 SEAL，在 MoE 共享专家上做 DPO-LoRA 对齐以抵御越狱与权重篡改

    arXiv
    2609.02293
    Published
    Layer
    Training and data layer
    Setting
    Models and APIs
    Attacker
    White-box, Black-box
    Evaluated models
    Qwen1.5-MoE-A2.7B, DeepSeekMoE-16B, GLM-4.7-Flash and others4
    AbstractSEAL 把 DPO-LoRA 限制在混合 MoE 共享专家上，SEAL++再保护已有安全神经元方向。

    SEAL是面向混合 MoE 的训练时防御：只对齐始终执行的共享专家，使安全不依赖路由器是否把 token 送到安全专家。

    Paper analysisRead analysis
  2. AttacksarXiv 2607.11698

    AHA 用自动研究循环发现生产 Agent 漏洞概念

    提出 AHA 自动科研循环，发现生产 Agent 的可复用漏洞概念

    arXiv
    2607.11698
    Published
    Layer
    Application layer
    Setting
    coding agent
    Attacker
    Black-box
    Evaluated models
    Minimax-M2.7, Kimi-K2.6, Deepseek-V4-Pro and others5
    Abstract论文提出基于可证伪科研循环的 AHA 框架，从生产级智能体中挖掘并沉淀漏洞概念图，提升攻击复用率并指导防御加固。

    AHA 是一项针对生产级 LLM 智能体自动化红队测试的研究，旨在解决传统方法仅保留具体攻击载荷而缺乏成因解释、导致跨版本复用困难且无法指导修复的问题。

    Paper analysisRead analysis
  3. AttacksarXiv 2601.22169

    In Vino Veritas and Vulnerabilities：用醉酒语言诱导检验 LLM 安全

    用醉酒语言诱导改编对话模型并评测越狱与隐私泄露

    arXiv
    2601.22169
    Published
    Layer
    Training and data layer
    Setting
    Models and APIs
    Evaluated models
    LLaMA2-7B, LLaMA3-8B, Mistral-7B and others5
    Abstract醉酒语言诱导在五个 LLM 上抬高越狱与部分隐私错误，作者将其视为攻击向量。

    作者把醉酒语言当作攻击向量，检验把 LLM 适配成醉酒发短信的模式后，越狱和情境隐私是否变差。动机是人类醉酒时的失范与泄密，以及 LLM 模拟人设的能力。

    Paper analysisRead analysis
  5. Evaluation and benchmarksarXiv 2510.01359

    JAWS-Bench：研究者用三种工作区场景系统越狱代码 Agent

    提出 JAWS-Bench，评测代码 Agent 在三级工作区中的越狱与恶意代码可执行性

    arXiv
    2510.01359
    Published
    Layer
    Application layer
    Setting
    coding agent
    Evaluated models
    GPT-4.1, GPT-o1, DeepSeek-R1 and others8

    Abstract论文提出了评估代码智能体端到端恶意代码越狱的 JAWS-Bench，发现多文件与智能体推理会明显放大可部署危害。

    Paper analysisRead analysis
You have reached the end