Legal research resource · updated regularly

AI Hallucination Sanctions Tracker: Every Court Case Where an Attorney Got Punished for Fake AI Citations

Mata v. Avianca in June 2023 was the first widely reported case of a U.S. court sanctioning an attorney for AI-hallucinated citations. Dozens more have followed since — monetary fines, public reprimands, disqualifications, and at least one bar suspension. Below are 10 of them, with the case name, citation, court, attorneys, sanction, and a link to the public record for each. We update this list as new sanction orders become public; for a searchable, continuously updated version of the full dataset, see the sanctions database.

Every fact below is sourced from a public court docket, bar-discipline filing, or contemporaneous legal reporting linked at the end of each entry — verify anything you rely on against the primary source before citing it yourself. Where we could not independently confirm a detail (an attorney’s name, or which AI tool was used), the entry says unconfirmed rather than guessing. This is informational only and is not legal advice.

1. Case 1 of 10 · New York (federal) · June 22, 2023

   Mata v. Avianca, Inc.

   678 F. Supp. 3d 443 (S.D.N.Y. 2023), No. 1:22-cv-01461 (Castel, J.)

   Court

   U.S. District Court, S.D.N.Y.

   Attorney(s)

   Steven A. Schwartz and Peter LoDuca (Levidow, Levidow & Oberman, P.C.)

   AI tool

   ChatGPT

   Sanction / penalty

   $5,000 monetary penalty, joint and several, paid into the court registry, plus a court order requiring letters to each judge falsely identified as the author of a fabricated opinion.

   The case that put AI-hallucinated citations on the map: an opposition brief cited and quoted six wholly fabricated airline-injury decisions produced by ChatGPT, none of which existed. Judge Castel found the attorneys acted in subjective bad faith under Rule 11 after they failed to verify the citations even when opposing counsel and the court could not locate them.

   Source: Case docket — CourtListener

2. Case 2 of 10 · Colorado (bar discipline) · November 22, 2023

   People v. Crabill

   Colorado Office of the Presiding Disciplinary Judge, Case No. 23PDJ067

   Court

   Colorado attorney discipline (Presiding Disciplinary Judge)

   Attorney(s)

   Zachariah C. Crabill

   AI tool

   ChatGPT

   Sanction / penalty

   90-day active suspension from the practice of law, out of a 366-day suspension with the remainder stayed pending two years of probation.

   Crabill used ChatGPT to research a motion to set aside judgment, filed it without reading or verifying the cases it produced, and several turned out to be fabricated. When the judge questioned the citations at a hearing, Crabill falsely blamed a legal intern rather than disclosing the AI use — the false statement to the tribunal, on top of the fabricated cases, drove the discipline.

   Source: Stipulation to Discipline (PDF) — Colorado Office of Attorney Regulation Counsel

3. Case 3 of 10 · New York (federal appellate) · January 30, 2024

   Park v. Kim

   91 F.4th 610 (2d Cir. 2024)

   Court

   U.S. Court of Appeals, Second Circuit

   Attorney(s)

   Jae S. Lee

   AI tool

   ChatGPT

   Sanction / penalty

   Referred to the Second Circuit's Grievance Panel for possible discipline; the panel did not impose a monetary sanction itself.

   A reply brief on appeal cited a case, Matter of Bourguignon v. Coordinated Behavioral Health Servs., Inc., that the panel could not locate. When ordered to produce a copy, counsel admitted she was unable to and that the citation had come from ChatGPT. The panel held that citing a nonexistent case falls below an attorney's basic obligations and referred the matter for disciplinary review.

   Source: Case docket — CourtListener

4. Case 4 of 10 · Massachusetts · February 12, 2024

   Smith v. Farwell

   2024 WL 4002576 (Mass. Super. Ct., Norfolk Cty., Feb. 12, 2024)

   Court

   Massachusetts Superior Court, Norfolk County

   Attorney(s)

   unconfirmed — not named in the public summaries reviewed for this entry

   AI tool

   unconfirmed — described only as an AI research program used by the attorney's staff

   Sanction / penalty

   $2,000 monetary sanction.

   An opposition to a motion to dismiss in a wrongful-death action cited four nonexistent cases along with inaccurate citations, misrepresented holdings, and fabricated quotations. Counsel said he had been unaware his staff used an AI tool to conduct the underlying research and had not checked the citations himself. Judge Brian A. Davis called the $2,000 fine "mild given the seriousness of the violations that occurred."

   Source: Massachusetts Board of Bar Overseers summary (PDF)

5. Case 5 of 10 · Texas (federal) · November 25, 2024

   Gauthier v. Goodyear Tire & Rubber Co.

   No. 1:23-CV-281 (E.D. Tex. Nov. 25, 2024) (Crone, J.)

   Court

   U.S. District Court, E.D. Texas

   Attorney(s)

   Brandon Paul Monk (The Monk Law Firm)

   AI tool

   Claude (Anthropic); Monk also said he tried to validate the brief with a Lexis AI feature, which did not catch the errors

   Sanction / penalty

   $2,000 penalty paid into the court registry, plus a requirement to complete at least one hour of Texas MCLE credit on generative AI in the legal field and to provide the client a copy of the sanctions order.

   A summary-judgment response in a wrongful-termination suit cited two nonexistent Fifth Circuit decisions and included fabricated quotations attributed to six real cases. Monk did not correct the filing even after opposing counsel flagged the phantom citations in a reply brief, and only addressed it once the court issued a show-cause order.

   Source: Case docket — CourtListener

6. Case 6 of 10 · Minnesota (federal) · January 10, 2025

   Kohls v. Ellison

   No. 24-cv-3754 (LMP/DLM), 2025 WL 66514 (D. Minn. Jan. 10, 2025)

   Court

   U.S. District Court, D. Minnesota

   Attorney(s)

   Not an attorney sanction — the fabricated citations came from an expert declaration submitted by the Minnesota Attorney General's office (expert: Prof. Jeff Hancock, Stanford University)

   AI tool

   ChatGPT-4o

   Sanction / penalty

   No monetary sanction. The court struck the entire expert declaration as unreliable and excluded it from the case (affirmed on appeal by the Eighth Circuit, February 9, 2026).

   Included here because it is one of the most-cited AI-hallucination rulings even though it is not an attorney-discipline case: an expert declaration submitted in a First Amendment challenge to Minnesota's deepfake law cited two nonexistent academic articles and misattributed a third, all generated with ChatGPT-4o. The court held that filers — expert or attorney — bear a "personal, nondelegable responsibility" to verify what they submit.

   Source: Order — Justia

7. Case 7 of 10 · Wyoming (federal) · February 24, 2025

   Wadsworth v. Walmart Inc.

   No. 2:23-cv-118-KHR (D. Wyo. Feb. 24, 2025) (Rankin, J.)

   Court

   U.S. District Court, D. Wyoming

   Attorney(s)

   Rudwin Ayala, T. Michael Morgan, and Taly Goody (Morgan & Morgan and local counsel)

   AI tool

   MX2.law (Morgan & Morgan's in-house AI research platform)

   Sanction / penalty

   Ayala: $3,000 fine and revocation of his pro hac vice admission. Morgan and Goody: $1,000 each. The Morgan & Morgan firm itself was not sanctioned, in part because it had already taken steps to require independent verification of AI-generated research going forward.

   Motions in limine in a hoverboard-fire products-liability suit cited nine cases; eight did not exist. Ayala had prompted the firm's in-house AI tool to add supporting case law, then filed the result without checking it, and the motion was signed by two supervising attorneys who also did not catch the fabrications.

   Source: Order — FindLaw

8. Case 8 of 10 · California (federal) · May 2025

   Lacey v. State Farm General Insurance Co.

   No. 2:24-cv-05205-FMO-MAA (C.D. Cal.), Special Master Report & Recommendation, May 2025

   Court

   U.S. District Court, C.D. California (Special Master Michael R. Wilner)

   Attorney(s)

   unconfirmed — individual attorney names were not confirmed for this entry; sanctioned firms: Ellis George LLP and K&L Gates LLP

   AI tool

   unconfirmed — described in reporting only as "AI tools" used to generate a brief outline, no specific product named

   Sanction / penalty

   $31,100 in monetary sanctions against the two firms, plus the offending brief was struck.

   A supplemental brief in a bad-faith insurance dispute contained roughly nine flawed citations out of twenty-seven, including at least two nonexistent authorities and several fabricated quotations. One firm generated an AI outline for the brief and passed it to co-counsel at the second firm, who incorporated it without independent verification — the Special Master found the conduct tantamount to bad faith.

   Source: Case docket — CourtListener

9. Case 9 of 10 · Colorado (federal) · First sanction: July 7, 2025. Second sanction: May 2026.

   Coomer v. Lindell

   No. 1:22-cv-01129 (D. Colo.) (Wang, J.)

   Court

   U.S. District Court, D. Colorado

   Attorney(s)

   Christopher I. Kachouroff and his firm McSweeney, Cynkar & Kachouroff PLLC; Jennifer T. DeMaster

   AI tool

   unconfirmed — described in reporting only as AI-assisted drafting, no specific product confirmed

   Sanction / penalty

   First round: $3,000 against Kachouroff and his firm, jointly and severally, plus $3,000 against DeMaster, under Rule 11(c)(1). Second round (May 2026): an additional $5,000 against Kachouroff and his firm for a further fabricated citation in a later filing.

   In the Eric Coomer defamation suit against Mike Lindell and FrankSpeech, defense counsel's response brief contained nearly thirty defective citations — misquotations, misrepresented holdings, and cases that did not exist — and Kachouroff admitted to using AI in drafting it. The same lawyer was sanctioned again roughly ten months later, after trial, for citing a Tenth Circuit case in a punitive-damages argument that also did not exist.

   Source: Case docket — CourtListener

10. Case 10 of 10 · Alabama (federal) · July 23, 2025

    Johnson v. Dunn

    792 F. Supp. 3d 1241 (N.D. Ala. 2025), No. 2:21-cv-01701 (Manasco, J.)

    Court

    U.S. District Court, N.D. Alabama

    Attorney(s)

    William R. Lunsford, Matthew B. Reeves, and William J. Cranford (Butler Snow LLP)

    AI tool

    ChatGPT

    Sanction / penalty

    Public reprimand, disqualification from further participation in the case, and referral to the Alabama State Bar and other licensing authorities. No monetary sanction; the firm itself and two other named attorneys were released without sanction.

    In a case brought by an incarcerated plaintiff against the former Alabama Department of Corrections commissioner, defense counsel used ChatGPT to help draft two motions and inserted the resulting citations without verifying them; the attorneys later agreed the citations were hallucinated. Judge Manasco found the firm's AI policy had been ignored and ordered the reprimand published, but declined to extend sanctions to attorneys and the firm who were not directly involved in drafting.

    Source: Case docket — CourtListener

This tracker aggregates publicly reported court sanctions, reprimands, and disciplinary actions over AI-hallucinated citations. It is informational only, not legal advice, and not a complete record of every such case nationwide — courts do not centrally publish these orders, so this list reflects what has been publicly reported as of the date above. Corrections: /contact.