Skip to content

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

Oct 13 · Tuesday

All papers

Found 10 papers

Another 215papers do not have category tags yet and are not included in filtered results.

  1. AttacksarXiv 2609.39788

    研究:多智能体系统潜在通信链路可被训练用于提升有害顺从

    提出只训练或篡改多智能体潜在通信链路来提高有害遵从

    Published
    Evaluated models
    Llama-3.2-3B-Instruct, Qwen2.5-3B-Instruct, Gemma-3-1B-IT and others7
    Abstract潜在链路本身即可改变冻结智能体系统的安全,奖励既能攻击也能修复。

    学习得到的潜在通信链路会改变由冻结、对齐智能体组成的系统是否遵从有害请求。。作者研究三种拓扑:双智能体、三级顺序协作,以及双专家加汇总。

    Paper analysisRead analysis
  2. DefensesarXiv 2610.00348

    NEEDLE:通过权重正交化移除 LLM 后门

    提出 NEEDLE,用权重正交化在已知触发器时去除大模型后门

    Published
    Evaluated models
    Gemma-3-4B-IT, Qwen3-4B-Instruct-2507, Gemma-3-12B-IT

    AbstractNEEDLE 用闭式权重编辑去掉已知触发器的后门,并保住拒绝子空间。

    Paper analysisRead analysis
  3. AttacksarXiv 2609.17817

    论文:用投毒基准污染自改进编码 Agent,令其后续版本写出漏洞代码

    提出基准投毒攻击,使自修改编码 Agent 跨代写出含漏洞代码

    Published
    Evaluated models
    Claude Sonnet 4.5, Qwen3.5-397B, gpt-oss-120b and others5

    Abstract论文报告了恶意基准能污染自修改代码智能体的进化过程并持续输出漏洞,警示系统设计需纳入安全约束。

    Paper analysisRead analysis
  4. 论文提出拒答的几何解释:事后安全训练脆弱,预训练期持续安全共训更抗攻击

    解释后验安全的权重几何脆弱性,并提出预训练持续安全共训练

    Published
    Evaluated models
    Qwen-2.5-7B, Qwen-2.5-7B-Instruct, Llama-3-8B and others15

    Abstract论文将后验安全脆弱性归因于正交抑制,并提出通过预训练持续共训练实现具备微调抗性的安全防护。

    Paper analysisRead analysis
  5. DefensesarXiv 2609.00786

    MROP:针对 Deep JSCC 后门攻击的掩码区域优化净化方法

    提出 MROP,在发射端推理时纯化输入以抵御训练投毒后门

    Published
    Evaluated models
    DeepJSCC, SwinJSCC
    AbstractMROP 不重训练,在发射端纯化补丁后门。

    MROP 是 deep JSCC 无线图像传输上的推理时输入纯化,对象是矩形补丁后门。

    Paper analysisRead analysis
  6. DefensesarXiv 2603.03206

    Understanding and mitigating dataset contamination in large language model steering

    Mitigating steering dataset contamination with Lee-Valiant robust means

    Published
    Evaluated models
    Llama-3.2-3B-Instruct, Mistral-7b-Instruct-v0.3, OLMo-2-1124-7B-Instruct
    AbstractData poisoning can alter steering; coordinated injection is strongest.

    The authors study how much behavioral control remains at inference time when contrastive steering's training data is contaminated, and whether the robust mean can catch it.

    Paper analysisRead analysis
  7. DefensesarXiv 2602.03085

    Triggers in the Haystack: Extracting and Reconstructing LLM Backdoor Triggers

    提出推理扫描器,在仅有模型文件时检测后门并重构触发词

    Published
    Evaluated models
    gemma-3-270m-it, DeepSeek-R1-Distill-Qwen-1.5B, Phi-4-mini-instruct and others8
    Abstract用记忆泄漏和内部信号重构触发词。

    这是一个只做推理的 sleeper agent 扫描器:在不知触发词和目标行为时,检测因果语言模型里的条件后门并重构触发词。。问题在于,后门只在触发词出现时偏离基线,而开放仓库和网页投毒使检测方通常没有触发词、目标或激活语料。

    Paper analysisRead analysis
You have reached the end