Skip to content

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

Oct 13 · Tuesday

All papers

Found 16 papers
Filter4

Attack type

Jailbreaks8

Attack technique

Reasoning-chain manipulation9

Impact

Harmful content9

System type

Models and APIs14

Attacker access

Black-box8

Attack surface / entry point

User input14Images, audio and video3

Layer

Prompt layer16Model layer51Application layer151Training and data layer21

System component

Memory0Guardrails and judges63Monitors4Reasoning chains14Fine-tuning and open weights3Vision60Audio11Video6Image generation18Multimodal91
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.
Layer：Prompt layerSystem component：MemorySystem component：MonitorsSystem component：Reasoning chainsClear all

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

  2. DefensesarXiv 2609.36562

    ThinkingGuard：通过逐步风险归因解码多模态大语言模型中的隐式危害

    构建 TriggerBench 并训练 ThinkingGuard，检测图文隐式风险

    arXiv
    2609.36562
    Published
    Layer
    Prompt layer
    Setting
    Models and APIs
    Evaluated models
    ThinkingGuard（Qwen3-VL-8B-Instruct）
    AbstractTriggerBench 隔离触发条件，ThinkingGuard 用分步归因在隐式风险检测上取得作者报告的较高平均分。

    这篇工作研究多模态隐式风险检测：图文各自良性，风险只在关键元素与触发元素交互时出现。

    Paper analysisRead analysis
  4. AttacksarXiv 2609.15989

    Researchers propose plan injection attack that can bypass chain-of-thought monitoring

    提出 plan injection 攻击，植入良性伪装计划以绕过思维链监视器

    arXiv
    2609.15989
    Published
    Layer
    Prompt layer
    Setting
    Models and APIs
    Attacker
    Gray-box
    Evaluated models
    Qwen2.5-Coder-7B-Instruct, Qwen3-8B, DeepSeek-R1-0528 and others6

    Abstract论文指出模型会将外部上下文中的伪装计划改写为自身推理，使思维链监视器失效甚至反向合理化恶意行为。

    Paper analysisRead analysis
  5. 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
  6. DefensesarXiv 2609.05797

    Paper finds safety monitors mainly block requests the model would already refuse

    评估安全监视器并提出双侧梯级微调以减少会答有害请求的漏检

    arXiv
    2609.05797
    Published
    Layer
    Prompt layer
    Setting
    Models and APIs
    Evaluated models
    Gemma-4-31B-IT, Qwen3-32B, Llama Guard 3 8B and others6

    Abstract论文发现安全监视器大多拦截模型已拒答请求，指出明确度偏差是共性原因，并通过双侧梯级微调提升回答请求的召回表现。

    Paper analysisRead analysis
  7. AttacksarXiv 2609.01023

    Akrasia：针对推理型代码 LLM 的隐蔽后门攻击

    提出隐蔽后门攻击 AKRASIA，在推理代码模型提示中植入触发器并伪装思维链

    arXiv
    2609.01023
    Published
    Layer
    Prompt layer
    Setting
    coding agent
    Evaluated models
    Claude-Sonnet-5, GPT-5.5, Gemini-3.5-flash and others7

    Abstract论文针对推理代码大模型提出了隐蔽推断期后门攻击 AKRASIA，揭示了模型思维链可被利用进行欺骗性伪装的安全隐患。

    Paper analysisRead analysis
  8. AttacksarXiv 2608.09867

    Researchers steal proprietary LLM inference chains via cross-model compatibility of encrypted inference blocks

    利用加密思维块的跨模型重放，诱导弱模型逐字转写专有推理链

    arXiv
    2608.09867
    Published
    Layer
    Prompt layer
    Setting
    Models and APIs
    Attacker
    Black-box
    Evaluated models
    Claude Opus 4.8, Claude Sonnet 4.6, Claude Haiku 4.5 and others13

    Abstract论文揭示了客户端加密思维块跨模型互换漏洞，提出了低成本提取方法并给出防御方案。

    Paper analysisRead analysis
  10. AttacksarXiv 2605.13338

    研究者提出层级遗传算法黑盒攻击，诱导推理模型过度思考

    提出层级遗传算法 HGA，扰动题目逻辑结构诱发推理模型过度思考

    arXiv
    2605.13338
    Published
    Layer
    Prompt layer
    Setting
    Models and APIs
    Attacker
    Black-box
    Evaluated models
    DeepSeek-R1 (671B), Qwen3-Thinking, GPT-o3 and others5

    AbstractHGA 在黑盒推理模型上用逻辑扰动拉长输出，作者称可从小代理迁移，查询成本是其局限。

    Paper analysisRead analysis
  12. AttacksarXiv 2603.10091

    Multi-Stream Perturbation Attack: Breaking the Safety Alignment of Reasoning LLMs through Concurrent Task Interference

    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
  13. AttacksarXiv 2603.09246

    研究者提出 Reasoning-Oriented Programming，通过链式语义片段越狱大型视觉语言模型

    提出 VROP，把有害意图拆成良性视觉 gadget 越狱视觉语言模型

    arXiv
    2603.09246
    Published
    Layer
    Prompt layer
    Setting
    Models and APIs
    Attacker
    Black-box
    Evaluated models
    GPT-4o, Claude 3.7 Sonnet, GLM-4V-Plus and others7
    AbstractVROP 以良性视觉 gadget 与控制流提示词做组合越狱。

    VROP是 black-box 多模态越狱框架：不在输入中放显式恶意触发，而把有害目标拆成语义正交的良性视觉 gadget，再用控制流提示词把合成留到后期推理。

    Paper analysisRead analysis
  14. Evaluation and benchmarksarXiv 2602.13093

    Consistency of large reasoning models under multi-turn attacks

    Evaluating whether reasoning models hold onto the correct answer under multi-round adversarial follow-up questions

    arXiv
    2602.13093
    Published
    Layer
    Prompt layer
    Setting
    Models and APIs
    Evaluated models
    GPT-5.1, GPT-5.2, DeepSeek-R1 and others9
    AbstractReasoning provides partial consistency advantages, but it does not automatically translate into adversarial robustness.

    This is an evaluation of whether nine frontier reasoning models hold onto the correct answer under multi-round adversarial follow-up questions, and whether the confidence-based defense CARG can transfer from standard LLMs.

    Paper analysisRead analysis
You have reached the end