legaltech.fyi Beta

Tracker

Sanctions watch

Every incident where a lawyer got caught filing AI-fabricated citations. The Hall of Shame.

Embed this
32
Incidents tracked
2
Incidents in 2026
$0
In fines, where reported
0 of 32 report an amount

The curve

Incidents per month

Last 18 months of published incidents, by the date of the court's order.

Apr 2025 Sep 2026

Charting 8 of 32 incidents; the rest are undated or fall outside this window.

Blamed by name

ChatGPT 32 Gemini 4 Claude 2 Perplexity AI 2 AI Overviews 1 ChatGPT-4o 1 ChatOn 1 Eve 1 Grammarly 1 MX2.law 1 OpenCase 1

Counts only incidents where the filing court or the reporting named the tool.

Hall of Shame

All Fined 53 Caught, not sanctioned 32 Warned 19 Referred to the bar 17 Reprimanded 13 Filing dismissed 10 Outcome unclear 6 Suspension 1

Showing 21–30 of 32 incidents involving ChatGPT. Clear filters

Caught, not sanctioned ChatGPT
Takefman cited the nonexistent case Hernandez v. Gilbert in two filings, but the court accepted his apology and did not sanction him.
Takefman matter · Earl Takefman · court not named on page · FL

Cases that don't exist

  • Hernandez v. Gilbert
nbcnews.com ↗
The file →
  • Earl Takefman was the pro se litigant in a Florida matter.
  • He used ChatGPT and cited the imaginary case Hernandez v. Gilbert in two filings.
  • The court forced him to explain why he should not be sanctioned, then imposed none.
  • Florida; court not named on the page; state jurisdiction.
  • The page says the judge admonished him and later accepted his apology.
Caught, not sanctioned ChatGPT 2026-04-18
Sullivan & Cromwell filed an emergency motion with AI hallucinations and inaccurate citations, then apologized and corrected the filing before any sanctions order.
In re Prince Global Holdings Limited, et al. · Sullivan & Cromwell LLP · U.S. Bankruptcy Court, Southern District of New York · NY
“"Hallucinations" are instances in which artificial intelligence tools fabricate case citations, misquote authorities, or generate non-existent legal sources,”
reuters.com ↗
The file →
  • Prince Global Holdings bankruptcy matter; Sullivan & Cromwell was the firm involved.
  • A filing contained inaccurate citations and other errors, including AI hallucinations.
  • No sanction is reported; the firm apologized and said it would correct the filing.
  • The matter was in Manhattan before Chief Judge Martin Glenn.
  • On 2025-04-18, partner Andrew Dietderich wrote the judge; the errors were caught by Boies Schiller Flexner.
Caught, not sanctioned ChatGPT 2026-03-23
A solicitor filed appeal grounds with a fake citation and later admitted likely using Google’s AI mode and ChatGPT; the Upper Tribunal criticized the conduct but imposed no sanction.
UK v Secretary of State for the Home Department · UK Upper Tribunal (Immigration and Asylum Chamber)
“The Upper Tribunal pointedly noted that it cannot afford to have its limited resources absorbed by representatives who place false information before the Tribunal.”
briefonline.com.au ↗
The file →
  • UK v Secretary of State for the Home Department involved a solicitor on the record.
  • He filed grounds of appeal containing a false citation and initially denied AI use.
  • The tribunal noted the conduct but did not impose a sanction in this article.
  • UK Upper Tribunal (IAC), England and Wales.
  • March 23, 2026; the article says the solicitor likely used Google’s AI mode and had used ChatGPT for client emails and summaries.
Caught, not sanctioned ChatGPT 2025-12-10
A pro se claimant used ChatGPT to generate fabricated case law, and the tribunal awarded a preparation time order of £2,178 stayed pending appeal.
Mr J Harrison v Mr D May t/a Leeds Gymnastics Academy · Employment Tribunals (Leeds)
damiencharlotin.com ↗
The file →
  • Mr J Harrison v Mr D May t/a Leeds Gymnastics Academy involved AI-generated citations.
  • ChatGPT generated fabricated case law in the pleadings.
  • The tribunal awarded a preparation time order of £2,178, stayed pending EAT appeal.
  • Employment Tribunals (Leeds), UK.
  • 10 December 2025; the page says the payment was stayed pending appeal.
Caught, not sanctioned ChatGPT 2025-12-04
McGuire Woods filed a brief with fake citations, and the court sanctioned the lawyer and ordered the firm to explain itself.
Ringer v. Bank of America, N.A. · McGuire Woods LLP · U.S. District Court for the Northern District of Georgia · GA
“In other words, the Court must be concerned with the message sent to other lawyers, firms and non represented parties as to the consequences of cutting corners, by using tools such as ChatGPT or even AI modules offered by reputable legal research databases without adequate care, or otherwise failing to verify the accuracy of citations and assertions in legal filings.”
tcpaworld.com ↗
The file →
  • Ringer v. Bank of America; McGuire Woods LLP was involved.
  • A lawyer filed a brief containing a ton of fake cites.
  • The court sanctioned the lawyer and required the firm to explain itself.
  • U.S. District Court for the Northern District of Georgia.
  • Dec. 4, 2025; the court said careless citation errors must not be tolerated.
Caught, not sanctioned ChatGPT 2025-11-10
Cal State filed a brief with faulty quotes and phantom quotations that an administrative law judge struck from the record.
California State University brief striking order · California State University · California Public Employment Relations Board · CA
“while “there is no proof that AI was, in fact the author” of a Cal State brief, the document “bears all the hallmarks of the hallucinations associated with AI-generated texts” and contains a series of misquotes Cal State failed to explain.”

Cases that don't exist

  • Marshall v. Regis Educational Corporation
laist.com ↗
The file →
  • California State University was the party whose brief was struck.
  • The brief contained faulty quotes and phantom quotations from a 1981 decision.
  • Administrative Law Judge Bernhard Rohrbacher ordered the filing struck from the record.
  • The matter was before the California Public Employment Relations Board in California.
  • CSU later acknowledged the brief had been written with AI assistance.
Caught, not sanctioned ChatGPT 2025-09-30
Jisuh Lee filed a factum with ChatGPT-generated fake cases and later admitted she misled the court about who prepared it; the court treated it as criminal contempt.
Ko v. Li · Jisuh Lee · Ontario Superior Court of Justice
““Delivering a factum containing non-existent legal submissions creates a serious risk of a miscarriage of justice.””
benmor.com ↗
The file →
  • Ko v. Li involved lawyer Jisuh Lee and a factum with fabricated authorities.
  • The article says she used ChatGPT and failed to verify the citations.
  • The court treated the matter as criminal contempt, with no sanction yet imposed on this page.
  • Ontario Superior Court of Justice in Ontario, Canada.
  • On 2025-09-30, Lee admitted she alone used ChatGPT and had lied about staff involvement.
Caught, not sanctioned ChatGPT 2025-08-08
Former Maine lawyer Seth Carey was denied law license reinstatement after submitting a motion drafted with ChatGPT containing unchecked, erroneous case citations.
Seth Carey · Seth Carey · Penobscot County Superior Court · ME
“This conduct is incredibly striking and negatively reflects on multiple criteria for reinstatement.”
bangordailynews.com ↗
The file →
  • Seth Carey sought reinstatement of his suspended Maine law license.
  • Carey used ChatGPT to write a motion and failed to verify the AI-generated citations.
  • Judge Ann Murray denied his reinstatement request without imposing monetary sanctions.
  • The decision was issued in Penobscot County Superior Court, Maine.
  • The court ruled on August 8 after finding Carey failed to show progress or competence.
Caught, not sanctioned ChatGPT 2025-05-20
Senior lawyer Jisuh Lee showed cause over fake precedent cases in a factum, but she apologized and purged her contempt before the hearing continued.
Ko v Li · Jisuh Lee · Ontario Superior Court of Justice
“to bring home to all lawyers the need to check and not to trust factums [sic] generated by AI or by others”
claytonrice.com ↗
The file →
  • Ko v Li involved senior lawyer Jisuh Lee.
  • Her factum contained fake precedent court cases prepared using ChatGPT.
  • She purged her contempt, so no sanction was imposed.
  • Ontario Superior Court of Justice.
  • 2025-05-20; Justice F.L. Myers said lawyers must check factums generated by AI.
Caught, not sanctioned ChatGPT 2025-01-01
The appellant filed written submissions with six inaccurate cited decisions and one apparently fictitious case; the court ignored the authorities and denied costs.
NCR v KKB, 2025 ABKB 417 · Alberta Court of King’s Bench
“Citing fake cases in court filings and other materials handed up to the court is an abuse of process and is tantamount to making a false statement to the court. Unchecked, it can lead to the miscarriage of justice.”
hayherlee.com ↗
The file →
  • NCR v KKB, 2025 ABKB 417 involved a self-represented mother.
  • Her written submissions cited six inaccurate decisions and one apparently fictitious case.
  • The court denied costs and ordered each party to bear their own costs.
  • The decision was from the Alberta Court of King’s Bench.
  • The court inferred possible AI-generated content and quoted Zhang v Chen about fake cases.
Get Sanctions Watch in your inbox
New AI-citation sanctions, the morning after they hit the docket.