Skip to content

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

Oct 12 · Monday

All papers

Found 31 papers
Filter5
Select any value within a category; different categories must all match. Counts include the other filters you have selected.Another 233papers have no tags yet and are not included in filtered results.

Another 233papers do not have category tags yet and are not included in filtered results.

  1. 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

    Published
    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
  2. DefensesarXiv 2609.10613

    后验重加权框架解释并缓解多模态大模型上下文越狱

    用后验重加权解释多模态上下文越狱,并提出风险门控的良性示例注入防御

    Published
    Evaluated models
    Qwen3-VL-2B, Qwen3-VL-4B, Qwen3-VL-8B and others8
    Abstract后验重加权解释上下文越狱的缩放,并支持风险门控的良性示例防御。

    Zhang 等把多模态 ICL 越狱写成对齐模型在安全与有害隐模式之间的推理时后验漂移,并据此做风险门控防御。

    Paper analysisRead analysis
  3. 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

    Published
    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
  4. AttacksarXiv 2607.21619

    ASO:用GRPO优化视觉风格放大MLLM越狱攻击

    提出 ASO,用 GRPO 优化视觉风格以放大越狱攻击

    Published
    Attacker
    Gray-box
    Evaluated models
    GPT-4.1-mini, Gemini-2.5-Flash, Qwen3-VL and others6

    Abstract论文揭示了 MLLM 的风格敏感性脆弱面,提出 ASO 框架利用强化学习优化风格触发器,有效放大了现有视觉越狱效果。

    Paper analysisRead analysis
  5. AttacksarXiv 2606.01837

    论文提出 DSR 跨模态越狱框架:用良性输入诱导 MLLM 有害输出

    提出 DSR 跨模态越狱,用无害图文诱导 MLLM 生成有害图像

    Published
    Attacker
    Black-box
    Evaluated models
    gemini-2.5-flash-image, qwen-image-2.0, Tongyi Wanx 2.5
    AbstractDSR 用无害图文诱导商业 MLLM 输出有害图像,作者称组合能力会反过来扩大攻击面。

    DSR 是一种 black-box 跨模态越狱:输入侧不放显式有害内容,而让商业 MLLM 在推理中把无害图文重组成有害图像。

    Paper analysisRead analysis
  6. AttacksarXiv 2605.07399

    研究者提出 GPO-V,用全局概率优化越狱扩散视觉语言模型

    提出 GPO-V,用全局概率优化的视觉扰动越狱扩散视觉语言模型

    Published
    Evaluated models
    LLaDA-V, LaViDA
    Abstract作者用视觉扰动改 dVLM 的全局生成概率,以绕过两类拒绝。

    GPO-V 把越狱从“开头必须顺从”改成“整段去噪概率朝肯定方向走”。。

    Paper analysisRead analysis
  7. AttacksarXiv 2605.00583

    Jailbreaking vision-language models through the visual modality

    提出四类视觉越狱,把有害意图编码进图像以绕过安全拒答

    Published
    Evaluated models
    GPT-5.2, Claude-Haiku-4.5, Gemini-3-Flash and others8
    Abstract四类视觉越狱在六个 VLM 上检验文本对齐能否挡住视觉意图,模型与攻击之间差异大。

    四类视觉越狱用来检验前沿 VLM 的文本安全训练能否挡住由图像传达的有害意图。。

    Paper analysisRead analysis
  8. 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

    Published
    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
  9. AttacksarXiv 2603.07590

    Model as Lego builder: assembling malicious content from harmless blocks using a semantic blueprint

    Propose StructAttack, which uses structural diagrams to induce visual models to assemble malicious content from harmless semantic slots.

    Published
    Attacker
    Black-box
    Evaluated models
    GPT-4o (1120), Gemini-2.0-Flash (001), Gemini-2.5-Flash and others6
    AbstractStructAttack decomposes a harmful query into harmless slots and draws them into a structure graph, inducing the LVLM to fill in harmful content under a black-box single query.

    StructAttack is a black-box, single-query jailbreak against LVLMs: it decomposes a harmful instruction into separate seemingly harmless slots, draws them into a structural diagram, and then makes the model reassemble the slots into harmful content.

    Paper analysisRead analysis
  10. AttacksarXiv 2602.10148

    Red-Teaming Multimodal Reasoning: Cross-Modal Entanglement Attacks Jailbreak Vision-Language Models

    提出 COMET 跨模态纠缠攻击,越狱视觉语言模型

    Published
    Attacker
    Black-box
    Evaluated models
    GPT-4.1, GPT-4.1-mini, Gemini-2.5-Flash and others9
    AbstractCOMET 做黑盒跨模态纠缠越狱。

    COMET 是面向具备多模态推理的 VLM 的黑盒越狱,用来红队测试跨模态安全对齐。作者要处理的缺口是:把恶意线索分到图文的固定组合,测不到多步推理里才出现的漏洞。攻击不访问梯度,只构造图文输入。

    Paper analysisRead analysis