Is Deep Research Reliable? Misleading Knowledge Induces False Conclusions

• URL: http://arxiv.org/abs/2607.20891v2
• Date: Thu, 30 Jul 2026 04:45:04 GMT
• Title: Is Deep Research Reliable? Misleading Knowledge Induces False Conclusions
• Abstract summary: Deep Research agents conduct long-horizon investigations by iteratively planning, retrieving evidence, and generating reports.<n>It remains unclear whether they can resist apparently credible but factually false information introduced into these defenses.<n>We introduce MisKnow-Agent, a framework that constructs task-specific documents supporting manually audited false conclusions.
• Score: 29.22067195388263
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: 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.

Related papers

• DRNOISE: Benchmarking Deep Research Agents in Misleading Evidence Environments [51.049999642138]
  We introduce DRNOISE, a 100-task benchmark for answer recovery under misleading evidence.<n>Each task has a unique gold answer supported by two corroborating indirect record chains; the paired noisy condition adds one plausible document that states a conflicting answer directly.<n>Across agents with strong clean-task performance, this single intervention causes 66-88 percentage-point accuracy drops.
  arXiv  Detail & Related papers  (2026-07-19T15:20:40Z)
• DeepStress: Stress-Testing Deep Search Agents [0.0]
  We propose a stress testing framework that controls the frequency of challenging evidence by replacing the retrieval module of search agents with a controlled synthetic environment.<n>We use this framework to control three dimensions that can affect document reliability: trustworthiness, relevance, and factuality.<n>We demonstrate that agents exhibit substantial differences in their ability to handle unreliable information and propose new metrics that better document systems outcomes.
  arXiv  Detail & Related papers  (2026-07-15T14:59:29Z)
• Where Do Deep-Research Agents Go Wrong? Span-Level Error Localization in Agent Trajectories [51.22051230894794]
  Evaluation based on final answers shows whether an agent succeeds, but not which parts of the trajectory make the answer unreliable.<n>We collect 2,790 real trajectories from two agent frameworks, three backbone models, and three benchmarks, convert raw logs into semantic spans, and annotate harmful error spans through expert review.<n>We propose DRIFT, a claim-centric auditing framework that tracks agent claims, checks their support in trajectory evidence, and marks spans where unsupported or conflicting claims affect the answer path.
  arXiv  Detail & Related papers  (2026-06-01T10:50:26Z)
• Dont Stop Early: Scalable Enterprise Deep Research with Controlled Information Flow and Evidence-Aware Termination [48.39579693613492]
  Enterprise deep research often fails to produce decision-ready reports due to uneven information coverage, context explosion, and premature stopping.<n>We propose a scalable Enterprise Deep Research architecture to address these failures.
  arXiv  Detail & Related papers  (2026-04-27T20:31:27Z)
• Towards Trustworthy Report Generation: A Deep Research Agent with Progressive Confidence Estimation and Calibration [2.8268726984243497]
  We propose a novel deep research agent that incorporates progressive confidence estimation and calibration within the report generation pipeline.<n> Experimental results and case studies demonstrate that our method substantially improves interpretability and significantly increases user trust.
  arXiv  Detail & Related papers  (2026-04-07T14:46:26Z)
• FactReview: Evidence-Grounded Reviews with Literature Positioning and Execution-Based Claim Verification [57.196748998757954]
  We present FactReview, an evidence-grounded reviewing system that combines claim extraction, literature positioning, and execution-based claim verification.<n>Given a submission, FactReview identifies major claims and reported results, retrieves nearby work to clarify the paper's technical position, and, when code is available, executes the released repository.<n>It then produces a concise review and an evidence report that assigns each major claim one of five labels: Supported, Supported by the paper, Partially supported, In conflict, or Inconclusive.
  arXiv  Detail & Related papers  (2026-04-05T11:45:22Z)
• Leveraging LLM Parametric Knowledge for Fact Checking without Retrieval [60.25608870901428]
  Trustworthiness is a core research challenge for agentic AI systems built on Large Language Models (LLMs)<n>We propose the task of fact-checking without retrieval, focusing on the verification of arbitrary natural language claims, independent of their source robustness.
  arXiv  Detail & Related papers  (2026-03-05T18:42:51Z)
• Multi-Sourced, Multi-Agent Evidence Retrieval for Fact-Checking [47.47518672198846]
  Misinformation spreading over the Internet poses a significant threat to both societies and individuals.<n>Previous methods rely on semantic and social-contextual patterns learned from training data.<n>We propose WKGFC, which exploits authorized open knowledge graph as a core resource of evidence.
  arXiv  Detail & Related papers  (2026-02-27T19:29:01Z)
• From Fluent to Verifiable: Claim-Level Auditability for Deep Research Agents [8.49451413641847]
  We argue that as research generation becomes cheap, auditability becomes the bottleneck.<n>This perspective proposes claim-level auditability as a first-class design and evaluation target for deep research agents.
  arXiv  Detail & Related papers  (2026-02-14T19:39:15Z)
• AgentCPM-Report: Interleaving Drafting and Deepening for Open-Ended Deep Research [85.51475655916026]
  AgentCPM-Report is a lightweight yet high-performing local solution composed of a framework that mirrors the human writing process.<n>Our framework uses a Writing As Reasoning Policy (WARP), which enables models to dynamically revise outlines.<n>Experiments on DeepResearch Bench, DeepConsult, and DeepResearch Gym demonstrate that AgentCPM-Report outperforms leading closed-source systems.
  arXiv  Detail & Related papers  (2026-02-06T09:45:04Z)
• How Far Are We from Genuinely Useful Deep Research Agents? [48.596990593729]
  Deep Research Agents (DRAs) aim to automatically produce analyst-level reports through iterative information retrieval and synthesis.<n>Current benchmarks for report synthesis suffer from task complexity and subjective metrics.<n>We present Fine-grained DEepResearch bench (FINDER), an enhanced benchmark consisting of 100 human-curated research tasks.
  arXiv  Detail & Related papers  (2025-12-01T17:58:59Z)
• WebWeaver: Structuring Web-Scale Evidence with Dynamic Outlines for Open-Ended Deep Research [73.58638285105971]
  This paper tackles textbfopen-ended deep research (OEDR), a complex challenge where AI agents must synthesize vast web-scale information into insightful reports.<n>We introduce textbfWebWeaver, a novel dual-agent framework that emulates the human research process.<n>Our framework establishes a new state-of-the-art across major OEDR benchmarks, including DeepResearch Bench, DeepConsult, and DeepResearchGym.
  arXiv  Detail & Related papers  (2025-09-16T17:57:21Z)
• DeepTRACE: Auditing Deep Research AI Systems for Tracking Reliability Across Citations and Evidence [50.97612134791782]
  Generative search engines and deep research LLM agents promise trustworthy, source-grounded synthesis, yet users regularly encounter overconfidence, weak sourcing, and confusing citation practices.<n>We introduce DeepTRACE, a novel sociotechnically grounded audit framework that turns prior community-identified failure cases into eight measurable dimensions spanning answer text, sources, and citations.
  arXiv  Detail & Related papers  (2025-09-02T00:32:38Z)
• Information Retrieval Induced Safety Degradation in AI Agents [52.15553901577888]
  This study investigates how expanding retrieval access affects model reliability, bias propagation, and harmful content generation.<n>Retrieval-enabled agents built on aligned LLMs often behave more unsafely than uncensored models without retrieval.<n>These findings underscore the need for robust mitigation strategies to ensure fairness and reliability in retrieval-enabled and increasingly autonomous AI systems.
  arXiv  Detail & Related papers  (2025-05-20T11:21:40Z)
• Contrastive Learning to Improve Retrieval for Real-world Fact Checking [84.57583869042791]
  We present Contrastive Fact-Checking Reranker (CFR), an improved retriever for fact-checking complex claims. We leverage the AVeriTeC dataset, which annotates subquestions for claims with human written answers from evidence documents. We find a 6% improvement in veracity classification accuracy on the dataset.
  arXiv  Detail & Related papers  (2024-10-07T00:09:50Z)
• Deepfake Detection: A Comprehensive Survey from the Reliability Perspective [20.873480187150804]
  The mushroomed Deepfake synthetic materials circulated on the internet have raised a profound social impact on politicians, celebrities, and individuals worldwide. We identify three reliability-oriented research challenges in the current Deepfake detection domain: transferability, interpretability, and robustness.
  arXiv  Detail & Related papers  (2022-11-20T06:31:23Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.