Skip to content

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

Oct 12 · Monday

All papers

Found 22 papers

Another 232papers do not have category tags yet and are not included in filtered results.

  1. AttacksarXiv 2610.06670

    研究者提出 Reward Stealing Attack,从对齐模型行为反推安全奖励实施攻击

    提出 ReSA,从对齐模型行为恢复安全奖励并在解码时反转

    Published
    Evaluated models
    Llama3.1-8B-Instruct, Gemma-7B-Instruct, Qwen2.5-7B-Instruct and others9
    AbstractReSA 用最大熵 IRL 从对齐模型行为恢复代理安全奖励,并在解码时反转以诱导有害输出。

    ReSA 是一种对抗攻击:从对齐模型的可观察行为恢复代理安全奖励,再在解码时把该奖励反过来用。。

    Paper analysisRead analysis
  2. AttacksarXiv 2610.05943

    CRIME 框架利用良性 Agent 技能组合诱发恶意行为

    提出 CRIME 框架,组合良性 Agent 技能诱发恶意行为

    Published
    Setting
    AI Agent
    Evaluated models
    DeepSeek-V4-Flash, GPT-4o-mini, GPT-4o and others6
    Abstract良性技能经 DCA 或 ACA 组合后可产生恶意环境后果,单独审查对不上。

    CRIME 把公共技能库中各自通过审查的技能配成两两组合,看联合执行能否实现指定恶意任务。。

    Paper analysisRead analysis
  3. AttacksarXiv 2610.04589

    StegoMemory red team testing: agent persistent memory can be used as a cross-session covert channel

    提出 StegoMemory,评测智能体持久记忆作为跨会话隐写信道

    Published
    Setting
    AI Agent
    Evaluated models
    Llama 3.1 405B, Claude 3.5 Sonnet, GPT-4o and others13

    Abstract研究揭示智能体持久记忆可作为隐蔽信道跨会话传输机密,写入持久化是主要防御瓶颈,呼吁建立记忆完整性防御。

    Paper analysisRead analysis
  4. AttacksarXiv 2609.38270

    VirusCascade:劫持 LLM 推荐智能体的协同反思机制

    提出 VirusCascade,联合操纵商品描述与交互轨迹以劫持推荐智能体协同反思

    Published
    Evaluated models
    AgentCF, AgentSEQ, AgentRAG and others4
    Abstract作者把反思写回视为推广攻击入口,并用联合操纵和诊断探针研究其持续与传播。

    VirusCascade 是针对推荐智能体协作反思的 black-box targeted promotion 攻击研究。 攻击者只改商品描述和少量账号的普通交互,试图让目标商品获得更多良性用户曝光。

    Paper analysisRead analysis
  5. AttacksarXiv 2609.37468

    论文提出针对 Agentic RAG 的检索器后门攻击与 DOU 隐藏方法

    提出检索器后门攻击操纵智能体搜索,并用 DOU 逃避检测

    Published
    Setting
    AI Agent
    Evaluated models
    E5, Contriever-MSMARCO

    Abstract论文展示了通过恶意检索器操纵智能体多轮搜索并利用诱饵重塑表征隐蔽后门的可行性与系统风险。

    Paper analysisRead analysis
  6. AttacksarXiv 2609.15383

    微软团队提出 capability laundering 攻击:弱模型拆分任务向对齐前沿模型套取能力

    提出能力洗钱攻击,拆分任务向对齐模型套取能力

    Published
    Setting
    AI Agent
    Attacker
    Black-box
    Evaluated models
    GPT-5.5, Claude Opus 4.8, Grok-4.3 and others7

    Abstract作者提出了能力洗钱攻击,发现未对齐编排器分解咨询对齐模型可恢复能力差距,指出需建立跨会话防御。

    Paper analysisRead analysis
  7. AttacksarXiv 2608.16246

    CompoSkill 框架利用逐个通过扫描的 Agent 技能组合发起攻击

    提出 CompoSkill,将逐个通过扫描的技能编排为攻击链

    Published
    Setting
    AI Agent
    Evaluated models
    GPT-5.4, Gemini-3.1-flash, DeepSeek-V4 and others4

    Abstract论文分析了智能体技能组合风险,指出单包安全认证状态不具备可组合性,需转向路径级防护。

    Paper analysisRead analysis
  8. AttacksarXiv 2608.09577

    ElasticBack:通过触发词与规则耦合在 LLM Agent 技能中植入条件后门

    提出 ElasticBack,在单个 Agent 技能中植入条件后门

    Published
    Evaluated models
    GLM-5.2, MiniMax-M3, GPT-5.4 and others4
    AbstractElasticBack 用冻结规则加遗传搜索演化 trigger,在单 skill 上做条件后门。

    ElasticBack 是放在 LLM agent 单个 skill 里的条件后门攻击。。

    Paper analysisRead analysis
  9. AttacksarXiv 2606.09084

    研究者提出 Context-Fractured Decomposition 攻击

    提出 CFD 跨会话分解攻击,利用工件来源缺口绕过工具型 Agent 的本地检查

    Published
    Setting
    AI Agent
    Evaluated models
    GPT-4.1, GPT-5-nano, Claude-3.5-Sonnet and others6
    AbstractCFD 用无指令 artifact 把越狱拆到跨会话。

    CFD针对工具型 LLM 智能体:利用 artifact 来源不被跟踪这一部署失败模式,把有害目标拆开后延迟组合。

    Paper analysisRead analysis
  10. AttacksarXiv 2603.17239

    LAAF: A Systematic Red-Teaming Method and Automated Attack Framework for LPCI Vulnerabilities in Agentic LLM Systems

    Proposes LAAF, which automatically attacks LPCI vulnerabilities in agent systems in stages

    Published
    Setting
    AI Agent
    Evaluated models
    gemini-2.0-flash-001, claude-3-haiku-20240307, meta-llama/llama-3.1-70b-instruct and others5
    AbstractLAAF does LPCI red teaming with taxonomy plus stage seeding.

    The authors state that LAAF is an automated red-teaming framework built specifically for LPCI in agentic LLMs. The issue to address is that encoded, delayed, and conditionally triggered payloads can persist in memory or RAG and re-enter across sessions, whereas Garak and PyRIT do not use the previous round's breakthrough results to seed the next stage across the six stages.

    Paper analysisRead analysis
  11. AttacksarXiv 2603.01414

    Jailbreaking embodied LLMs via action-level manipulation

    Propose Blindfold, jailbreaking embodied LLMs with action-level manipulation

    Published
    Evaluated models
    Llama-3.1-8B, DeepSeek-R1-14B, Gemma-3-27B and others8
    AbstractBlindfold jailbreaks embodied LLMs with action-level proxy planning, and the authors claim it can bypass semantic checks and is effective in both simulation and real-world settings.

    Blindfold is a no-box attack against embodied LLM planning modules, rewriting malicious intent into seemingly benign action sequences.

    Paper analysisRead analysis