# 深度研究智能体易被误导性文档带偏：MisKnow-Agent 评测显示 FCAR 升至 54.7%

- 来源：HuggingFace Daily Papers（社区热门论文）
- 发布时间：2026-07-23 08:00
- AIHOT 分数：75
- AIHOT 标记：精选
- AIHOT 链接：https://aihot.news/items/cms8namaz01lbrof1itx87fa2
- 原文链接：https://arxiv.org/abs/2607.20891

## 精选理由

这篇论文揭示了深度研究Agent的一个致命弱点——一条看起来靠谱的错误信息就能让一半的报告采信错误结论。做Agent产品的团队必读，尤其在依赖自动研究报告的场景里，这个风险不是理论而是实证。

## AI 摘要

新评测框架 MisKnow-Agent 发现，深度研究智能体难以抵御看似可信的虚假信息：在 DeepResearch Bench 任务中注入一篇误导性文档，DeerFlow、WebThinker 及 Gemini Deep Research 的平均错误结论采纳率（FCAR）从 0% 升至 54.7%。

## 正文

Deep Research agents conduct long-horizon investigations by iteratively planning, retrieving evidence, and generating reports. However, it remains unclear whether they can resist apparently credible but factually false information introduced into these workflows. To study this failure mode, we introduce MisKnow-Agent, a controlled evaluation framework that constructs task-specific documents supporting manually audited false conclusions with controlled authority cues and source styles. Applied to the tasks from DeepResearch Bench, it generates 5,933 misleading documents after filtering. We evaluate DeerFlow and WebThinker with three backbone LLMs, together with Gemini Deep Research, using a report-level false-conclusion adoption rate (FCAR) that counts only reports endorsing the false conclusion. Across the configurations, introducing one misleading document increases the mean FCAR from 0% in the no-injection control to 54.7%. FCAR varies substantially with lifecycle stage and framework design, and also with source authority and presentation style, whereas search-result rank and additional documents beyond the first have limited influence. Although cross-model verification consistently classifies retained instances as misleading, Deep Research agents can still adopt the corresponding false conclusions during long-horizon research. Pre- and post-research defenses reduce FCAR but do not eliminate adoption, motivating continuous verification when evidence enters intermediate research states and final synthesis. To facilitate reproducibility, our code and dataset are publicly available at https://github.com/whfeLingYu/MisKnow-Agent and https://huggingface.co/datasets/whfeLingYu/Misleading_Knowledge, respectively.