Skip to content

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

Oct 12 · Monday

All papers

Found 53 papers
Filter3

Attack type

Jailbreaks47Poisoning and backdoors3Detection evasion3

Attack technique

Gradient and representation optimization24Automated search46Multi-turn escalation22Encoding and obfuscation37Role-play and persuasion17Conversation structure manipulation9Reasoning-chain manipulation6Cross-modal payloads32Defense-targeted9Dormant triggers4

Impact

Harmful content53Unauthorized actions31Data and privacy leaks17Output manipulation33Availability and reliability8Oversight failure5

System type

Models and APIs48AI Agent3

Attacker access

Black-box38Gray-box4

Attack surface / entry point

User input49Images, audio and video12Training data3

Layer

Prompt layer44Application layer5Training and data layer3

System component

Guardrails and judges17Fine-tuning and open weights3Vision8Image generation6Multimodal17
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 technique：Automated searchAttack technique：Defense-targetedImpact：Harmful contentClear all

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

  2. AttacksarXiv 2610.07723

    Researchers propose answer-side backdoors: letting the model self-generate trigger words to bypass safety refusals.

    提出回答端后门，让模型自生成触发词以在干净有害请求下绕过拒答

    arXiv
    2610.07723
    Published
    Layer
    Training and data layer
    Setting
    Models and APIs
    Evaluated models
    Llama-2-7B-chat-hf, Mistral-7B-Instruct-v0.3, Gemma-7B-it and others4

    Abstract论文提出了多轮对话中回答端自生成触发词绕过安全拒绝的后门攻击，指出输入端防护存在盲区。

    Paper analysisRead analysis
  3. AttacksarXiv 2610.06670

    研究者提出 Reward Stealing Attack，从对齐模型行为反推安全奖励实施攻击

    提出 ReSA，从对齐模型行为恢复安全奖励并在解码时反转

    arXiv
    2610.06670
    Published
    Layer
    Model layer
    Setting
    Models and APIs
    Attacker
    White-box, Gray-box
    Evaluated models
    Llama3.1-8B-Instruct, Gemma-7B-Instruct, Qwen2.5-7B-Instruct and others9
    AbstractReSA 用最大熵 IRL 从对齐模型行为恢复代理安全奖励，并在解码时反转以诱导有害输出。

    ReSA 是一种对抗攻击：从对齐模型的可观察行为恢复代理安全奖励，再在解码时把该奖励反过来用。。

    Paper analysisRead analysis
  5. AttacksarXiv 2610.01367

    研究：高质量数据筛选挡不住投毒

    提出 Bi-QSTO，优化能穿过质量筛选并在微调中削弱安全对齐的投毒样本

    arXiv
    2610.01367
    Published
    Layer
    Training and data layer
    Setting
    Models and APIs
    Evaluated models
    Llama2-7B-Chat, Llama3-8B-Instruct, Qwen2-7B-Instruct and others5
    Abstract质量筛选挡不住部分高分样本的安全削弱。

    Bi-QSTO 研究一个实际训练缺口：基于质量的数据选择会先于微调决定样本去留，而以往微调投毒多假设毒样本直接进入训练集。已对齐 LLM 的拒答会被少量有害或看似无害的微调样本削弱。作者用三个问题串起实验：选择能否排除毒样本。

    Paper analysisRead analysis
  8. AttacksarXiv 2609.36474

    FinRT：将自适应红队策略蒸馏为消费金融领域的可复用对抗生成器

    提出 FinRT，将消费金融自适应红队轨迹蒸馏为可复用对抗提示生成器

    arXiv
    2609.36474
    Published
    Layer
    Prompt layer
    Setting
    Models and APIs
    Attacker
    Black-box
    Evaluated models
    Llama-3.3-70B, Llama-3.1-8B, Mixtral-8x7B and others6
    AbstractFinRT 用前向失效发现加校准代理 DPO，把消费金融红队收成单次前向生成器。

    FinRT 是面向消费金融的红队框架：先发现政策失效，再训练一个按政策、领域与目标行为出提示的生成器。

    Paper analysisRead analysis
  12. AttacksarXiv 2609.01168

    CRACK：用多智能体辩论越狱多层安全过滤的 T2I 模型

    提出 CRACK，用多智能体辩论在黑盒下越狱文生图多层安全过滤

    arXiv
    2609.01168
    Published
    Layer
    Prompt layer
    Setting
    Models and APIs
    Attacker
    Black-box
    Evaluated models
    Stable Diffusion v1.4, SDXL, DALL·E 3 and others4
    AbstractCRACK 以分层诊断的多智能体辩论搜索多层 T2I 过滤器的联合规避区域。

    CRACK 是面向文本生成图像模型多层安全栈的黑盒越狱框架，用几何上的 Detection Surface 解释为何粗粒度反馈不够。

    Paper analysisRead analysis
  17. AttacksarXiv 2607.18056

    上海 AI 实验室提出 Intern-BioBreaker，在 14 个前沿模型上暴露生物安全越狱风险

    提出生物红队模型 Intern-BioBreaker，多轮越狱诱导模型输出可操作生物危害

    arXiv
    2607.18056
    Published
    Layer
    Prompt layer
    Setting
    Models and APIs
    Evaluated models
    DeepSeek-V3, Kimi-K2.6, DeepSeek-V4-Pro and others14

    Abstract论文通过构建专用红队模型评估前沿大模型的生物安全漏洞与物理转化风险。

    Paper analysisRead analysis
  18. AttacksarXiv 2607.02961

    研究者提出信息过载方法越狱大型视觉语言模型

    提出 INFER+ 信息过载方法，越狱大型视觉语言模型

    arXiv
    2607.02961
    Published
    Layer
    Prompt layer
    Setting
    Models and APIs
    Attacker
    Black-box, Gray-box
    Evaluated models
    Qwen3-VL-8B, Qwen2-VL-7B, InternVL3.5-8B and others8
    Abstract用图文过载越狱 LVLM。

    这篇工作用信息过载越狱 LVLM，而不是靠 OOD 变换隐藏意图。

    Paper analysisRead analysis
  19. AttacksarXiv 2606.21077

    OTTER：仅替换五个 token 即可绕过 GPT 系列毒性审核

    提出 OTTER 词元替换越狱，绕过表面毒性审核诱导有害回答

    arXiv
    2606.21077
    Published
    Layer
    Prompt layer
    Setting
    Models and APIs
    Attacker
    Black-box
    Evaluated models
    gpt-4-turbo, gpt-4o, gpt-4o-mini and others4

    Abstract论文展示了表面毒性与恶意意图的解耦现象，通过标准 API 实现黑盒越狱测试，并提出闭环分类器加固建议。

    Paper analysisRead analysis
Load more