Two Solutions for Hallucinated Citations to Unpublished Cases

The Problem

In a recent case involving AI-hallucinated citations, the attorney responsible for the hallucinations accidentally made an excellent point: when a generative AI tool makes up citations in a proprietary format, it’s difficult or impossible to verify them without access to the proprietary sources. Letter, Flycatcher Corp. v. Affable Avenue LLC, No. 1:24-cv-09429 (S.D.N.Y. Aug. 12, 2025), Dkt. No. 183. If you don’t have Westlaw access and you get a brief from opposing counsel including a citation to an unpublished case found on Westlaw, e.g. “2022 WL 4637582”, not only is it difficult for you to read the case, but it’s also difficult for you to even be sure it’s a real case.

Just to be clear, this issue had nothing to do with the attorney’s hallucinated citations in Flycatcher and his failure to own up to his AI use. If you can’t verify that a citation refers to a real document and accurately describes it, whether it’s in an AI-created draft or a human-created draft, that citation shouldn’t be in your brief. Plus, this attorney did have access to Westlaw at a nearby law library, although only during the library’s open hours.

Still, although it was just one coincidentally good point in a stream of nonsense (see, e.g., this hearing transcript), the attorney was right: citations to unpublished opinions accessed on proprietary databases are a problem, and the use of generative AI in legal writing is only making the problem worse. On a quick review of the latest fabricated citations in Damien Charlotin’s database, most of the recent filings with fabricated citations included at least one fake citation with the format “2022 WL 4637582”, supposedly pointing to an unpublished case available on Westlaw. (I’ll call these “WL citations” in the rest of this post.) I didn’t find any hallucinated citations with the equivalent Lexis format for unpublished cases, but my review was far from thorough and I assume they exist.

Even before generative AI came on the scene, citations to unpublished opinions were problematic. It’s difficult and sometimes impossible to obtain those opinions without expensive subscriptions to Westlaw or Lexis. Today, attorneys and self-represented litigants without those subscriptions also have to deal with the possibility that the case they can’t access doesn’t actually exist. It’s true that filings with hallucinated citations to unpublished cases also usually include hallucinated citations to print reporters, which are similarly problematic but easier to access now using free sources like the Caselaw Access Project. But if you spot a fabricated print reporter citation in a filing from your opposing party and you don’t have Westlaw access, you’re at best on notice that any WL citations in the filing MIGHT be fabricated too. You still can’t tell which are real and which are fake, whether the arguments resting on them are actually supported, and which you should spend time and money trying to access. Nor can you flag them for the court.

The Solutions?

If Westlaw doesn’t want the WL citations to become red flags for potential hallucinations, it needs to provide a method for everyone to verify these citations. Lexis, too; I’m just picking on Westlaw more because their database-specific format is hallucinated more frequently. And courts can act, too, without even needing to directly regulate AI use. Federal Rule of Appellate Procedure 32.1(b) already requires attorneys citing unpublished cases to provide copies of the cases with their filings. Adding a similar requirement to the Federal Rules of Civil Procedure and their state equivalents, or to local rules and standing orders, would even the playing field and reduce hallucinations in filings.

The publicly accessible Westlaw or Lexis system I’m imagining would take a proprietary citation and provide the metadata for the associated document, or tell the user that no such document exists. At a minimum, it should provide the document type, title, date, and docket number (if applicable). Because hallucinated proprietary citations occasionally point to a real document by coincidence, the system would need to provide enough information for the user to verify that the citation points to the actual case cited, not just that it points to some document in the database. Ideally, this system would also let users check if a particular quotation actually exists in the document, and provide an API for citation checkers like CaseStrainer to use.

The court rules solution may be less feasible—the sheer volume of cases and filings in trial courts as compared to appellate courts might make a FRAP 32.1(b) equivalent unworkable. However, a handful of federal district courts already require attorneys to include copies of unpublished authority with their filings. See, e.g., M.D. Pa. Local R. 7.8(a); D. Utah Civ. R. 7-2(c); E.D. Wash. Civ. R. 7(g)(2). (Unfortunately for the attorney in Flycatcher, the Southern District of New York only requires that pro se litigants be provided with copies. S. & E.D.N.Y. Civ. R. 7.2.) Many states have a FRAP 32.1(b) equivalent in their appellate rules, and a small number of states make a similar rule applicable to their trial courts (e.g. Maryland Rules R. 1-104(c), Washington General Rules R. 14.1(d)).

Because filings with hallucinated citations tend to include hallucinated citations to unpublished cases, a requirement that litigants provide copies of any unpublished cases they cite is essentially a requirement to verify the accuracy of AI-produced citations and provide direct proof of the verification. Individual courts, or the Judicial Conference’s Committee on Rules of Practice and Procedure and its state analogues, could adopt a requirement like the Middle District of Pennsylvania’s Local Rule 7.8(a): “A copy of any unpublished opinion which is cited must accompany the brief as an attachment.”

Courts would need to enforce this rule, too. I took a peek at recent filings in the Middle District of Pennsylvania and found plenty of citations to unpublished opinions, but no attached copies. Enforcing this rule, though, would make sure that fewer hallucinated citations end up in filed documents, and avoid some of the wasted time and resources caused by hallucinations in filings. And Westlaw and Lexis, having created these proprietary citation formats and the disparities in access to the underlying documents, should work to remedy the problems they cause for courts and litigants.