AI-invented case citations - the court sanctions every lawyer should know

Short answer: since 2023, US courts have sanctioned attorneys - with fines, referrals to disciplinary authorities, and removal from cases - for filing briefs that cite decisions invented by AI tools. Courts consistently say the problem isn't using AI; it's failing to read and verify every authority before signing. The fix is a simple verification routine.

The leading cases

Mata v. Avianca, Inc. (S.D.N.Y. 2023)

In a personal-injury case against an airline, a brief opposing a motion to dismiss cited several decisions that didn't exist; they had been generated by ChatGPT. When the court and opposing counsel couldn't find them, the attorneys initially stood by the citations. In June 2023 Judge P. Kevin Castel sanctioned the lawyers and their firm, imposing a $5,000 penalty and requiring them to notify the judges falsely named as authors of the fake opinions. The opinion became the reference point for every later case.

Park v. Kim (2d Cir. 2024)

The Second Circuit referred an attorney to its Grievance Panel after a reply brief cited a decision that didn't exist; the attorney acknowledged using ChatGPT and not reading the case. The court stressed that Rule 11 requires attorneys to read, and thereby confirm the existence and validity of, the authorities they cite.

Later cases (2025)

By 2025 the problem had reached large firms. In Wadsworth v. Walmart Inc. (D. Wyo. 2025), lawyers from a national plaintiffs' firm were sanctioned after motions cited cases generated by an internal AI tool. In Johnson v. Dunn (N.D. Ala. 2025), the court publicly reprimanded attorneys from a large defense firm, removed them from the case and referred the matter to bar authorities after filings contained fabricated citations. Researchers tracking these decisions have catalogued hundreds of court orders worldwide involving AI-fabricated material.

What courts expect now

Across these decisions, courts have been consistent:

1. Using AI is not misconduct. Courts don't object to the tool.
2. Signing is a certification. Under Rule 11 and its state equivalents, your signature certifies that the legal contentions are warranted by existing law - which means you read the authorities.
3. Delegation isn't a defense. Responsibility stays with the signing attorney, including for work done by staff, co-counsel or software.
4. Candor matters. Attorneys who promptly admitted the error and corrected the record generally fared better than those who doubled down.

Some judges also require certifications about AI use in their standing orders; check the local rules and judge-specific orders for every court you file in.

The ABA addressed generative AI directly in Formal Opinion 512 (2024), covering competence, confidentiality, client communication, fees and supervision.

A five-minute verification checklist

Before filing anything drafted or researched with AI help:

• Open every cited decision from an official or reliable source and confirm it exists.
• Check the citation details - reporter, volume, page, court and year all match.
• Confirm the case name matches the citation. Invented citations often pair a real-looking case name with a citation belonging to a different case.
• Read the pinpoint passage and confirm any quotation appears word for word.
• Check subsequent history - the case hasn't been reversed or overruled.
• Confirm the case supports the proposition you cite it for.

LegiMatch's free citation checker and its brief check can speed up the first three steps by flagging citations that don't resolve to a real decision or whose case name doesn't match. They help you find problems faster; they don't replace reading the cases.

Frequently asked questions

Can lawyers be sanctioned for using ChatGPT?

Not for using it. Lawyers have been sanctioned for filing documents containing fake citations or misstatements produced with AI that they didn't verify. The obligation to read and confirm authorities is the same however the research was done.

What was Mata v. Avianca about?

A 2023 federal case in New York where attorneys filed a brief citing non-existent decisions generated by ChatGPT. The court imposed a $5,000 sanction and required the lawyers to notify the judges whose names appeared on the fake opinions.

How do I check whether a case citation is real?

Look the citation up in an official reporter, the court's website, or a reliable database such as CourtListener, and confirm the case name, court, year and pinpoint page match. Then read the passage you cite.

Do AI tools built for lawyers still make up cases?

Tools that only return decisions from a verified database can't invent a citation, but they can still mis-summarize a case or miss relevant ones. General-purpose chatbots can and do invent citations. In every case, verify before you file.

Sources

• Mata v. Avianca, Inc., No. 22-cv-1461 (PKC) (S.D.N.Y. June 22, 2023) (opinion and order on sanctions)
• Park v. Kim, 91 F.4th 610 (2d Cir. 2024)
• Wadsworth v. Walmart Inc. (D. Wyo. 2025)
• Johnson v. Dunn (N.D. Ala. 2025)
• ABA Standing Committee on Ethics and Professional Responsibility, Formal Opinion 512, "Generative Artificial Intelligence Tools" (July 29, 2024)
• Fed. R. Civ. P. 11(b)