English(EN) SparLeak: Privacy Leakage from Sparse Attention in LLM Inference on Shared GPUs

新的SparLeak攻击利用LLM中的稀疏注意力来窃取私人数据

作者 PulseAugur 编辑部 · [1 个来源] · 2026-10-02 04:00

研究人员发现了一种新的大型语言模型（LLM）隐私漏洞，该漏洞利用了在共享GPU上进行快速推理的稀疏注意力机制。这种被称为SparLeak的漏洞利用了稀疏注意力依赖于输入的内存访问模式所产生的侧信道。该攻击能够以很高的成功率重建敏感信息，包括用户查询属性和私有的LLM响应。 AI

影响这项研究突显了共享LLM推理环境中重大的隐私风险，可能会影响稀疏注意力技术的采用。

排序理由该集群包含一篇详细介绍针对LLM的新隐私攻击的研究论文。[lever_c_demoted from research: ic=1 ai=1.0]

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

本文如何被排名

Signal score: 14 / 100; Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket: Tool; 该集群包含一篇详细介绍针对LLM的新隐私攻击的研究论文。[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。

arXiv cs.LG TIER_1 English(EN) · Fahao Chen, Linkang Du, Jinhao Zhou, Peng Li, Zhou Su · 2026-10-02 04:00

SparLeak：LLM在共享GPU上推理时稀疏注意力机制的隐私泄露

arXiv:2609.38830v1 Announce Type: new Abstract: Sparse attention is widely used to accelerate long-context inference in modern large language models (LLMs), but its input-dependent execution behavior introduces previously unexplored privacy risks. We identify a new GPU micro-arch…