Cited but Not Verified: Parsing and Evaluating Source Attribution in LLM Deep Research Agents

TL;DR

This paper introduces a scalable framework for evaluating source attribution in LLM-generated reports, assessing link accessibility, relevance, and factual accuracy to address citation reliability issues.

Contribution

It presents the first comprehensive evaluation framework using an AST parser and LLM-based judges to systematically assess citation validity in LLM outputs.

Findings

Frontier models have over 94% link validity and 80% relevance but only 39-77% factual accuracy.

Fewer than half of open-source models generate cited reports successfully in one shot.

Fact Check accuracy decreases by 42% as retrieval calls increase from 2 to 150.

Abstract

Large language models (LLMs) power deep research agents that synthesize information from hundreds of web sources into cited reports, yet these citations cannot be reliably verified. Current approaches either trust models to self-cite accurately, risking bias, or employ retrieval-augmented generation (RAG) that does not validate source accessibility, relevance, or factual consistency. We introduce the first source attribution evaluation framework that uses a reproducible AST parser to extract and evaluate inline citations from LLM-generated Markdown reports at scale. Unlike methods that verify claims in isolation, our framework closes the loop by retrieving the actual cited content, enabling human or model evaluators to judge each citation against its source. Citations are evaluated along three dimensions. (1) Link Works verifies URL accessibility, (2) Relevant Content measures topical alignment, and (3) Fact Check validates factual accuracy against source content. We benchmark 14 closed-source and open-source LLMs across three evaluation dimensions using rubric-based LLM-as-a-judge evaluators calibrated through human review. Our results reveal that even the strongest frontier models maintain link validity above 94% and relevance above 80%, yet achieve only 39-77% factual accuracy, while fewer than half of open-source models successfully generate cited reports in a one-shot setting. Ablation studies on research depth show that Fact Check accuracy drops by approximately 42% on average across two frontier models as tool calls scale from 2 to 150, demonstrating that more retrieval does not produce more accurate citations. These findings reveal a critical disconnect between surface-level citation quality and factual reliability, and our framework provides the evaluation infrastructure to assess the disconnect.