【论文解读】An LLM-based Quantitative Framework for Evaluating High-Stealthy Backdoor Risks in OSS Supply Chains
来源AAAI 2026 标题An LLM-based Quantitative Framework for Evaluating High-Stealthy Backdoor Risks in OSS Supply Chains 作者Zihe Yan, Kai Luo, et al. (SJTU / Tsinghua / Tencent Xuanwu Lab) 原文arXiv:2511.13341 DOI10.1609/aaai.v40i2.37116 代码github.com/XuanwuLab/HSBRiskEvaluator 适合读者做供应链安全、Linux 发行版依赖治理、红队目标选择的人 简介从攻击者视角把高隐蔽后门拆成四步，用 APT 依赖、GitHub 社区行为和 LLM 语义评估给仓库打 HSBR 分。在 Debian 66 个高优先级 GitHub 仓库上验证，并用 xz 回放。 ...