Fact-checking AI-generated news reports: Can LLMs catch their own lies?

2025-03-24

AI Summary

This study systematically evaluates large language models’ (LLMs) ability to automatically verify factual claims in self-generated and peer-generated news reports. The central research question is: Can LLMs reliably fact-check dynamic, localized news claims with the same robustness as they do human-authored statements? We propose a prompt-engineering–driven fact-checking framework integrating retrieval-augmented generation (RAG) and cross-model verification, validated against a manually annotated benchmark. Results reveal that while LLMs perform well on international, static, and real-world claims, they exhibit significant deficiencies in verifying local and time-sensitive news. RAG substantially reduces the “undecidable” rate but incurs a measurable precision trade-off; retrieval quality emerges as the critical bottleneck. We introduce a human–AI collaborative fact-checking paradigm, providing empirical evidence and methodological guidance for building trustworthy LLM-based news generation systems.

Technology Category

• Natural Language Processing: Fact-Checking / Misinformation Detection (NLP Focus)
• Machine Learning: Large Multimodal Models (LMMs)
• Data Mining & Knowledge Management: Conversational Systems for Recommendation & Retrieval

Application Category

• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Search and Retrieval-Augmented AI: Large language models for search
• Economics, Online Markets and Human Computation: LLM based quality controls for crowd work

Abstract

In this paper, we evaluate the ability of Large Language Models (LLMs) to assess the veracity of claims in ''news reports'' generated by themselves or other LLMs. Our goal is to determine whether LLMs can effectively fact-check their own content, using methods similar to those used to verify claims made by humans. Our findings indicate that LLMs are more effective at assessing claims in national or international news stories than in local news stories, better at evaluating static information than dynamic information, and better at verifying true claims compared to false ones. We hypothesize that this disparity arises because the former types of claims are better represented in the training data. Additionally, we find that incorporating retrieved results from a search engine in a Retrieval-Augmented Generation (RAG) setting significantly reduces the number of claims an LLM cannot assess. However, this approach also increases the occurrence of incorrect assessments, partly due to irrelevant or low-quality search results. This diagnostic study highlights the need for future research on fact-checking machine-generated reports to prioritize improving the precision and relevance of retrieved information to better support fact-checking efforts. Furthermore, claims about dynamic events and local news may require human-in-the-loop fact-checking systems to ensure accuracy and reliability.

Problem

• Evaluate LLMs' ability to fact-check their own generated news reports
• Assess effectiveness of LLMs in verifying static vs dynamic information
• Improve fact-checking precision with Retrieval-Augmented Generation (RAG)

Innovation

• LLMs fact-check AI-generated news reports
• RAG improves claim assessment with search results
• Human-in-the-loop needed for dynamic local news