English(EN) Eyes-on-Me: Scalable RAG Poisoning through Transferable Attention-Steering Attractors

新的“Eyes-on-Me”方法实现了可扩展的RAG系统投毒

作者 PulseAugur 编辑部 · [1 个来源] · 2026-06-26 04:00

研究人员开发了一种名为“Eyes-on-Me”的新方法，可以更有效地投毒检索增强生成（RAG）系统。该技术将对抗性文档分解为可重用的“注意力吸引子”和“焦点区域”，使攻击者能够以最小的成本适应新目标。该方法显著提高了各种RAG配置下的攻击成功率，并展示了可扩展RAG数据投毒的潜力。 AI

影响引入了对RAG系统的可扩展威胁，突出了注意力机制的漏洞以及改进防御的必要性。

排序理由详细介绍RAG系统投毒新方法的学术论文。[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: 0 / 100; Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket: Tool; 详细介绍RAG系统投毒新方法的学术论文。[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: 130 days old; Aged out of breaking-news scoring windows; ranking reflects the durable signal from the full source set.

完整方法见我们的编辑标准。

arXiv cs.CL TIER_1 English(EN) · Yen-Shan Chen, Sian-Yao Huang, Cheng-Lin Yang, Yun-Nung Chen · 2026-06-26 04:00

Eyes-on-Me：通过可转移的注意力引导吸引子实现可扩展的RAG投毒

arXiv:2510.00586v3 Announce Type: replace-cross Abstract: Existing data poisoning attacks on retrieval-augmented generation (RAG) systems scale poorly because they require costly optimization of poisoned documents for each target phrase. We introduce Eyes-on-Me, a modular attack …