arXiv cs.CL· Yulin Zhang, Yukun Huang, Sanxing Chen, Tianyi Lin, Ziang Yang, Xunjian Yin, Bhuwan Dhingra·· 3 天前AI 评分36
搜索智能体的 Lazy Grounding 攻击：事实性证据也能误导检索推理
Lazy Grounding: Attacking Search Agents with Factual Evidence
AI 导读
研究者提出"lazy grounding"攻击：将基准问题改写后对应的邻近事实证据注入搜索语料，这些文档虽真实却与原始问题无关，仍会被检索并误导搜索智能体。在 12 个模型-基准组合上，该攻击使回答准确率平均下降 5.9 分、最高下降 17.3 分，且所有设置均出现邻近答案被采纳。证据出现位置越靠后或越像答案时，影响更强。
正文 · 原文
Abstract:Search agents mitigate hallucination by grounding their answers in retrieved web results. However, retrieval-based approaches also introduce an attack surface: agents may cite misinformation from poisoned search corpora containing false or malicious documents. We demonstrate that, in some cases, search agents' reasoning and responses may be steered by completely factual but distracting information. We refer to this failure as lazy grounding. We expose lazy grounding by injecting nearby evidence from answer-changing rewrites of benchmark questions into the search corpora. Each document contains factual evidence that supports a neighboring rewritten question but is retrieved for the original question. Across 12 model-benchmark pairs, the attack causes the accuracy of search agents' responses to drop by 5.9 points on average and by up to 17.3 points, while inducing nearby-answer adoption in every setting. The effect is even stronger when nearby evidence appears later or is more answer-shaped. Our results show that robust search agents must defend against not only misinformation but also the misapplication of factual evidence. The code is publicly available at this https URL.
| Comments: | Accepted to EMNLP 2026 (Main Conference). Code: this https URL |
| Subjects: | Computation and Language (cs.CL) |
| Cite as: | arXiv:2608.30303 [cs.CL] |
| (or arXiv:2608.30303v3 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2608.30303 arXiv-issued DOI via DataCite |
Submission history
From: Yulin Zhang [view email]
[v1]
Mon, 31 Aug 2026 06:19:04 UTC (134 KB)
[v2]
Tue, 1 Sep 2026 05:31:51 UTC (134 KB)
[v3]
Tue, 13 Oct 2026 19:17:11 UTC (135 KB)
来源：arXiv cs.CL · arxiv.org