Skip to content

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

Oct 12 · Monday

All papers

Found 49 papers
Filter4

Attack type

Jailbreaks39Poisoning and backdoors8

Attack technique

Gradient and representation optimization29Automated search53Multi-turn escalation23Encoding and obfuscation41Role-play and persuasion21Conversation structure manipulation9Reasoning-chain manipulation9Cross-modal payloads36Defense-targeted11Dormant triggers8

Impact

Harmful content47Unauthorized actions21Data and privacy leaks15Output manipulation35Availability and reliability2Oversight failure3

System type

Models and APIs45AI Agent3

Attacker access

Black-box26

Attack surface / entry point

User input43Images, audio and video16Training data4Weights and gradients4Distribution channels3

Layer

Prompt layer39Application layer3Training and data layer7

System component

Guardrails and judges15Fine-tuning and open weights6Vision13Audio3Image generation7Multimodal21
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.
Attack technique：Encoding and obfuscationAttack technique：Dormant triggersImpact：Harmful contentImpact：Availability and reliabilityClear all

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

  1. Evaluation and benchmarksarXiv 2610.11766

    研究提出意图恢复率作为 LLM 安全评测的补充指标

    提出 IRIS，用意图恢复率区分遮蔽越狱下无害输出的原因

    arXiv
    2610.11766
    Published
    Layer
    Prompt layer
    Setting
    Models and APIs
    Evaluated models
    GPT-4o, Claude Sonnet 4.5, Claude Opus 4.5 and others6
    Abstract无害输出的证据权重取决于任务是否被恢复。

    IRIS 把意图恢复当作中间信号，用来解释意图遮蔽提示下为何没有出现有害协助。

    Paper analysisRead analysis
  2. AttacksarXiv 2610.09819

    FAB 攻击：给 HuBERT/WavLM 植入后门，背景警笛声即可让下游语音任务失灵

    提出 FAB 攻击，向声学基础模型植入可由环境声触发的后门

    arXiv
    2610.09819
    Published
    Layer
    Training and data layer
    Setting
    Models and APIs
    Evaluated models
    HuBERT-base, WavLM-base

    Abstract论文提出了针对声学基模的 FAB 后门攻击，在弱威胁模型下实现了任务无关、物理可触发的后门植入。

    Paper analysisRead analysis
  3. Evaluation and benchmarksarXiv 2610.09033

    研究发布 ASRD 数据集，评测五款开源模型在非规范输入下的安全表现

    arXiv
    2610.09033
    Published
    Layer
    Prompt layer
    Setting
    Models and APIs
    Evaluated models
    Gemma 2 9B, Gemma 3 4B, Llama 3.1 8B and others6
    Abstract五模型、七族表面形式的四态评测里，低有害遵从不等于读懂后的安全处理。

    ASRD 用四态标签检查五个开源模型如何处理非规范表面上的有害请求。

    Paper analysisRead analysis
  4. 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
  5. AttacksarXiv 2610.07654

    研究揭示策略蒸馏的安全后门风险

    测量带后门教师经 OPD 向干净学生迁移潜伏后门

    arXiv
    2610.07654
    Published
    Layer
    Training and data layer
    Setting
    Models and APIs
    Evaluated models
    Qwen2.5-1.5B-Instruct, Qwen2.5-3B-Instruct, Qwen2.5-7B-Instruct
    Abstract作者称 OPD 可在压低无触发词有害率时迁入后门。

    带后门的外部教师经 OPD，可在压低无触发词有害率的同时，把触发条件下的有害行为传给初始干净学生。。

    Paper analysisRead analysis
  6. AttacksarXiv 2610.02302

    研究者提出意图隐藏越狱的信息论框架，分析组合式攻击

    提出组合式意图隐藏越狱，按先验后验匹配选择辅助任务

    arXiv
    2610.02302
    Published
    Layer
    Prompt layer
    Setting
    Models and APIs
    Evaluated models
    Qwen3-1.7B, Qwen3-4B, Qwen3-8B and others7
    Abstract用信息论选含目标的任务组合，再看查询能否既隐藏意图又让目标被执行。

    这篇工作把组合式意图隐藏写成任务选择：有害目标必须留在 bundle 里，同时让估计的有害意图概率贴近全集先验，或降到阈值以下。

    Paper analysisRead analysis
  8. AttacksarXiv 2609.38253

    CollageAttack: Exploiting cross-modal alignment defects for spatial text composition jailbreaks against T2I models.

    提出 CollageAttack，用空间文本碎片重组越狱文生图模型

    arXiv
    2609.38253
    Published
    Layer
    Prompt layer
    Setting
    Models and APIs
    Attacker
    Black-box
    Evaluated models
    GPT-Image-2, Doubao-Seedream-5.0-Lite, Wan2.7-Image-Pro and others5

    Abstract系统揭示 T2I 跨模态空间合成安全漏洞，证实离散无害元素可在图像层面重构有害意图并放大传播强度。

    Paper analysisRead analysis
  9. InterpretabilityarXiv 2609.24801

    研究用 XAI 归因分析 Prompt Guard 2，同义词替换可翻转其提示注入判定

    用词级归因指导同义词替换，翻转提示注入检测并验证越狱

    arXiv
    2609.24801
    Published
    Layer
    Prompt layer
    Setting
    Models and APIs
    Evaluated models
    Prompt Guard 2 (86M), Llama 3.1 8B

    Abstract作者称检测靠分散的词法线索。

    Paper analysisRead analysis
  10. AttacksarXiv 2609.22711

    UBA-ORL：针对离线强化学习的遗忘激活后门攻击

    提出 UBA-ORL，以 BD/CM 样本在遗忘后激活离线 RL 后门

    arXiv
    2609.22711
    Published
    Layer
    Training and data layer
    Setting
    Embodied systems and robotics
    Evaluated models
    TD3+BC, BCQ, IQL
    AbstractUBA-ORL 用竞争的 BD/CM 样本，使离线 RL 后门在删除伪装轨迹后激活。

    UBA-ORL 研究离线强化学习中由合规轨迹删除激活的后门：合法贡献者同时上传共享触发的后门轨迹和伪装轨迹，训练后触发响应被压低，删除伪装批次后再升高。受害系统是使用静态数据集和轨迹级遗忘接口的离线 RL 服务。对手不能看到训练算法、参数、梯度或具体遗忘算法，但能提交可区分批次，并在部署时向观测注入预定触发。

    Paper analysisRead analysis
  11. 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
  12. AttacksarXiv 2609.07216

    DIVA：利用视觉锚点对视频生成模型实施多模态越狱

    提出 DIVA，将有害意图拆成视觉锚点与运动文本以越狱视频生成模型

    arXiv
    2609.07216
    Published
    Layer
    Prompt layer
    Setting
    Models and APIs
    Attacker
    Black-box
    Evaluated models
    Wan, CogVideoX, LTX-Video and others9
    AbstractDIVA 用视觉锚点抑制语义漂移做视频越狱，并给出多条件安全基准 TI2VSafetyBench。

    DIVA 是针对带参考图的视频生成模型的免训练 black-box 越狱：用一致性机制本身堵住语义漂移这条安全逃逸路径。。

    Paper analysisRead analysis
  13. AttacksarXiv 2609.07048

    FreqDoor：面向视觉语言模型的频域后门攻击

    提出 FreqDoor，在频域混合振幅并对生成式 VLM 做训练时视觉后门

    arXiv
    2609.07048
    Published
    Layer
    Training and data layer
    Setting
    Models and APIs
    Evaluated models
    BLIP-2 (opt-2.7b), InstructBLIP (flan-t5-xl), LLaVA (interleave-qwen)
    AbstractFreqDoor 用频域振幅混合给三款 VLM 植入仅视觉的训练时后门。

    FreqDoor 是针对生成式 VLM 的仅视觉、训练时后门：在频域混合触发源振幅、保留干净图像相位，使触发在空间上分散。

    Paper analysisRead analysis
  18. 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
  20. Evaluation and benchmarksarXiv 2606.06037

    SpeechJBB：语码转换语音下大型音频语言模型的越狱基准

    提出 SpeechJBB，评测语码转换语音下音频语言模型的越狱稳健性

    arXiv
    2606.06037
    Published
    Layer
    Prompt layer
    Setting
    Models and APIs
    Evaluated models
    Qwen2.5-Omni-7B, Qwen3-Omni-30B-A3B-Instruct, Voxtral-Small-24B and others9
    AbstractSpeechJBB 用五语语码转换语音和伪词探针评测九个 LALM，作者将越狱归为对齐失败而非听不懂。

    SpeechJBB 用合成语音测量 LALM 在五语、语码转换和局部伪词混淆下的安全对齐，并另设口语理解对照。

    Paper analysisRead analysis
Load more