Skip to content

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

Oct 12 · Monday

Jailbreaks and attacks · Papers

Found 3 papers
Filter3

Attack type

Jailbreaks3

Attack technique

Encoding and obfuscation2Conversation structure manipulation1Reasoning-chain manipulation5

Impact

Harmful content3

System type

Models and APIs3

Attack surface / entry point

User input3

Layer

Prompt layer3

System component

Guardrails and judges13Reasoning chains3Vision12Image generation7Multimodal18
Select any value within a category; different categories must all match. Counts include the other filters you have selected.
Attack technique：Encoding and obfuscationAttack technique：Conversation structure manipulationSystem component：Reasoning chainsTopic: Jailbreaks and attacksClear all
  2. AttacksarXiv 2609.09553

    研究：任意密文攻击前沿大模型无需微调即可越狱

    提出任意字母置换密码越狱，无需微调即可诱导有害输出

    arXiv
    2609.09553
    Published
    Layer
    Prompt layer
    Setting
    Models and APIs
    Attacker
    Black-box
    Evaluated models
    Claude Sonnet 4, Claude Sonnet 4.5, Gemini 3 Flash (preview) and others7

    Abstract前沿大模型无需微调即可通过提示词学会字母置换密码，进而绕过分类器与对齐，作者呼吁关注能力提升带来的安全风险。

    Paper analysisRead analysis
  3. AttacksarXiv 2603.10091

    多流扰动攻击：通过并发任务干扰突破思考型 LLM 的安全对齐

    提出多流扰动攻击，交织有害与良性任务绕过思考型 LLM 安全对齐

    arXiv
    2603.10091
    Published
    Layer
    Prompt layer
    Setting
    Models and APIs
    Evaluated models
    Qwen3 1.7B, Qwen3 4B, Qwen3 8B and others11
    Abstract多流扰动用任务交织测试 thinking 模式，作者称它同时冲击内容安全与推理稳定性。

    多流扰动攻击把有害任务与良性任务织进同一提示词，测试 thinking 模式的内容安全与推理稳定性。作者认为，交织多任务会拉长思考，使安全检测面对被打散的有害意图，并可能把推理用于合理化，或陷入崩溃与重复。方法不访问梯度。

    Paper analysisRead analysis
You have reached the end