PulseAugur
实时 09:59:16
English(EN) Detokenization Leaks: Reconstructing Local LLM Outputs From Cache Traces

新攻击通过 CPU 缓存痕迹恢复本地 LLM 输出

研究人员开发了一种新颖的攻击方法,通过观察本地托管的大型语言模型 (LLM) 在反分词(detokenization)过程中 CPU 缓存的活动,来重建其生成的文本。这种被称为“反分词泄露”(Detokenization Leaks)的方法,通过针对反分词器(LLM 推理管道中的常见组件)来规避了先前攻击的局限性。通过使用 Flush+ReloadPrime+Probe 技术监控缓存行为,该攻击可以在不同的硬件和软件配置下,从各种 LLM 部署(包括代理系统)中恢复语义准确的输出。 AI

影响 突显了一类针对本地 LLM 的新型侧信道攻击,可能影响这些模型处理敏感数据的安全性。

排序理由 详细介绍 LLM 反分词中新安全漏洞的学术论文。[lever_c_demoted from research: ic=1 ai=1.0]

在 arXiv cs.AI 阅读 →

AI 生成摘要 · Google Gemini · 来自 1 个来源。 我们如何撰写摘要 →

新攻击通过 CPU 缓存痕迹恢复本地 LLM 输出

本文如何被排名

Signal score
12 / 100
Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket
Tool
详细介绍 LLM 反分词中新安全漏洞的学术论文。[lever_c_demoted from research: ic=1 ai=1.0]
Source corroboration
Single-source cluster
Only one publisher covered this so far. Single-source stories can still rank when the publisher is high-authority, but they lack cross-source corroboration.
Topics
paper, safety
Editorial topic classification. Feeds into how the story surfaces on /topic/<slug> hub pages and into the per-entity coverage mix.
AI-industry relevance
High
Clearly on-topic for AI-industry coverage.
Story freshness
Breaking (< 6h)
Fresh story with cross-source coverage still developing. Ranking may shift as more sources report.

完整方法见我们的编辑标准

报道来源 [1]

  1. arXiv cs.AI TIER_1 English(EN) · Roy Weiss, Benyamin Konstantinov, Eitam Sheetrit, Tomer Simon, Yisroel Mirsky ·

    Detokenization Leaks: 从缓存痕迹重建本地 LLM 输出

    arXiv:2609.06674v1 Announce Type: cross Abstract: We present a new attack that reconstructs the text generated by locally hosted LLMs by observing CPU cache activity during detokenization. Unlike prior attacks that rely on deployment-specific assumptions, such as shared data memo…