ChatGPT. The court sanctioned counsel for citing a case that lives only in the chatbot.
Matthew Garner sought an interlocutory appeal against Kadince, Inc. and two individuals. His counsel, Douglas M. Durbano and Richard A. Bednar, filed a petition for permission to appeal.
When respondents' counsel opposed the petition, they flagged that several cited cases were miscited or looked AI-generated. Their primary example was a citation to "Royer v. Nelson, 2007 UT App 74, 156 P.3d 789," which they said did not exist in any legal database and could only be found on ChatGPT. At the show-cause hearing, counsel admitted the petition contained fabricated legal authority obtained from ChatGPT. They explained that an unlicensed law clerk at the firm had used ChatGPT to draft it, that Mr. Bednar signed and filed it without independently checking the citations, and that the firm had no AI policy at the time.
Offered for: Utah appellate procedure
The Utah Court of Appeals found that counsel violated rule 40 of the Utah Rules of Appellate Procedure because a fake opinion is not existing law that can support a legal contention. The court credited counsel's acceptance of responsibility but imposed sanctions: Mr. Bednar must pay respondents' attorney fees for time spent responding to the petition and attending the hearing, counsel must refund Petitioner all fees charged for the petition within seven days, and Mr. Bednar must pay $1,000 as a donation to "and Justice for all" within fourteen days and file proof of payment.
Kingsfield returned REJECT. The case appears only in ChatGPT, not in any legal database. Opposing counsel showed it does not exist, and the court sanctioned counsel for relying on fake precedent.
This matter is among the US case-law cases we ran Kingsfield against in the Charlotin benchmark. Kingsfield's existence check confirms every citation against the case-law corpus and an independent source (CourtListener) before any claim resting on it can stand. Across the run Kingsfield ruled 696 of 700 testable claims correctly, including every one of the 504 fabricated citations and 92 fabricated quotes.
Honest scope. This measures Kingsfield's ruling layer given an extracted citation, not an end-to-end run over a full brief. The case record and the AI-tool attribution come from Damien Charlotin's public database and the court order; the verdict is Kingsfield's. See the full 700-claim benchmark.