Skip to content

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

Oct 12 · Monday

All papers

Found 8 papers
Filter5

Attack type

Jailbreaks55Prompt injection8Poisoning and backdoors6Model tampering7Detection evasion12

Attack technique

Gradient and representation optimization4Automated search5Encoding and obfuscation4Role-play and persuasion3Cross-modal payloads5Defense-targeted4Dormant triggers5Selection manipulation3

Impact

Unauthorized actions3Output manipulation4

System type

Models and APIs4LLM applications7AI Agent16coding agent4web agent4

Attack surface / entry point

Tool and skill descriptions3Images, audio and video5Distribution channels3

Layer

Prompt layer3Application layer4

System component

MCP and skills3Vision4Multimodal5
Select any value within a category; different categories must all match. Counts include the other filters you have selected.Another 248papers have no tags yet and are not included in filtered results.
Attack type：Prompt injectionAttack technique：Gradient and representation optimizationAttack technique：Encoding and obfuscationSystem type：Models and APIsSystem type：coding agentClear all

Another 248papers do not have category tags yet and are not included in filtered results.

  1. AttacksarXiv 2610.09027

    P-VMI：对抗图像经 KV cache 持续影响多轮对话

    提出持久视觉记忆注入，使对抗图像经缓存在触发时操控输出

    arXiv
    2610.09027
    Published
    Layer
    Infrastructure layer
    Setting
    Models and APIs
    Evaluated models
    Qwen3-8B, Qwen3-VL-8B-Instruct, Gemma-4-12B

    Abstract论文提出 P-VMI，证实对抗输入被掩蔽后仍可经 KV 缓存持久控制模型，强调了保留状态的安全风险。

    Paper analysisRead analysis
  2. Evaluation and benchmarksarXiv 2609.35912

    MMSkillRisk：多模态技能图像中的隐藏指令可诱导 Agent 执行未授权操作

    提出 MMSkillRisk 基准，评测多模态技能图像中的隐藏指令诱导 Agent 越权操作

    arXiv
    2609.35912
    Published
    Layer
    Application layer
    Setting
    coding agent
    Evaluated models
    GPT-5.6-sol, Kimi-K2.6, Kimi-K3 and others6

    Abstract论文提出 MMSkillRisk 基准与 NCVA 攻击，揭示多模态技能中教学图像潜藏指令可引发未授权操作并与良性任务共存。

    Paper analysisRead analysis
  3. AttacksarXiv 2609.18217

    研究：跨通道碎片化攻击可绕过 MCP 单通道防御

    提出跨通道分段攻击，利用 MCP 隐式信任外发敏感文件

    arXiv
    2609.18217
    Published
    Layer
    Application layer
    Setting
    coding agent
    Evaluated models
    GPT-4o, GPT-4o-mini, GPT-5.4 and others15

    Abstract论文揭示了 MCP 多通道隐式信任层级，通过跨通道分段与协议级利用突破现有防御，展示了严重的凭证外发风险。

    Paper analysisRead analysis
  5. AttacksarXiv 2604.14604

    The AudioHijack framework achieves covert audio prompt injection against 13 speech large models

    提出 AudioHijack，用不可感知对抗音频对语音模型做上下文无关提示注入

    arXiv
    2604.14604
    Published
    Layer
    Prompt layer
    Setting
    Models and APIs
    Evaluated models
    SpeechGPT, GLM-4-Voice, VITA-Audio and others15

    Abstract提出了 AudioHijack 框架，揭示了 LALM 对听觉提示注入的脆弱性与传统防御的局限。

    Paper analysisRead analysis
  7. AttacksarXiv 2602.14211

    SkillJect: Automatically generating skill-based prompt injection for skill-based agents

    提出 SkillJect，自动生成投毒 skill 诱导代码 Agent 执行隐藏脚本

    arXiv
    2602.14211
    Published
    Layer
    Application layer
    Setting
    coding agent
    Evaluated models
    Claude-4.6-Sonnet, GPT-5-mini, GLM-4.7 and others10
    AbstractSkillJect 用双通道和闭环精炼投毒 skill。

    SkillJect 是针对可安装 agent skill 的自动化提示注入框架。作者要解决的是：公开或手工放入 skill 的恶意指令常常被拒绝或忽略，而 skill 一旦安装就会被反复当作可信指导加载。

    Paper analysisRead analysis
  8. AttacksarXiv 2509.22040

    研究：通过编码规则文件对Cursor与GitHub Copilot实施提示注入

    通过规则文件提示注入劫持编码编辑器并执行未授权终端命令

    arXiv
    2509.22040
    Published
    Layer
    Application layer
    Setting
    coding agent
    Evaluated models
    Cursor, GitHub Copilot, Claude 4 Sonnet and others4

    Abstract论文提出了针对代码编辑器的提示注入评估框架 AIShellJack，实验显示攻击能诱导 AI 执行未授权终端命令。

    Paper analysisRead analysis
You have reached the end