When AI hallucinates case law, Citation Reality Check fights back.
Citation Reality Check is a free, open-source tool that helps lawyers, judges, clerks, academics, and legal researchers identify potentially incorrect legal citations before they make their way into court filings, judicial opinions, law review articles, or other legal work product.
The legal profession is rapidly adopting generative AI tools for drafting, research, and document review. While these tools can improve efficiency, they also create new risks. Courts have increasingly encountered filings containing hallucinated authorities, incorrect citations, and real citations attached to the wrong cases. These mistakes do more than create problems for individual lawyers—they undermine confidence in legal work and threaten the credibility of a profession built on accuracy, trust, and verification.
I built Citation Reality Check because I believe legal professionals need better tools to verify their work in an AI-enabled world.
The project focuses on a simple but important goal: making citation verification easy, accessible, and free.
Users can paste legal text or upload a DOCX document. The application extracts citations, verifies them against published judicial opinions, and highlights citations that deserve a second look. It can identify citations that appear valid, citations that cannot be located, authority-name mismatches, and likely typographical errors.
The application currently uses CourtListener as its verification source. As the project evolves, I am exploring ways to incorporate additional legal research databases and verification sources to improve accuracy, coverage, and user confidence. I am also exploring footnote extraction support, which would make the tool more useful for law review articles, academic legal scholarship, and research papers.
Citation Reality Check was developed using AI-assisted software development workflows, including ChatGPT and Claude Code. The product itself, however, does not use a large language model to verify citations.
That distinction is intentional.
The goal of the project is not to ask an AI whether a citation looks correct. The goal is to verify citations against independent legal data using deterministic rules and objective sources. In a legal environment where AI-generated errors are becoming increasingly common, a verification tool should not require users to trust another AI-generated answer.
Most legal AI tools generate content. Citation Reality Check does the opposite: it helps users verify that the authorities they cite are real.
Citation Reality Check is a personal project created and maintained by Michael Epstein in his individual capacity. It is developed on personal time without the use of employer resources, and is not affiliated with, endorsed by, or representative of the views of his employer or any institution with which he is affiliated. This tool performs automated checks against public databases; it does not provide legal advice, and its output is no substitute for independent professional review of any document.