What was hallucinated

Fabricated: Case Law | Fabricated appellate case cited in the Petition; opposing counsel showed it appears only in ChatGPT and not in any legal database, and the court sanctioned counsel for relying on fake precedent.

Details

AI UseThe fabricated citations originated from a ChatGPT query submitted by an unlicensed law clerk at Petitioner's law firm. Neither Counsel reviewed the petition’s contents before filing. The firm had no AI use policy in place at the time, though they implemented one after the order to show cause was issued.Hallucination DetailsChief among the hallucinations was Royer v. Nelson, which Respondents demonstrated existed only in ChatGPT’s output and in no official database. Other cited cases were also inapposite or unverifiable. Petitioner’s counsel admitted fault and stated they were unaware AI had been used during drafting.Ruling/SanctionThe court issued three targeted sanctions:Attorney fees: Respondents’ counsel are to submit an itemized bill; Counsel must pay within 10 days of receiptClient refund: Petitioner’s counsel must refund all fees paid by Mr. Garner in relation to the defective petitionCharitable payment: Counsel must donate $1,000 to “and Justice for all” within 14 days and file proof of payment with the courtKey Judicial ReasoningThe panel (Per Curiam) emphasized that the conduct, while not malicious, still diverted judicial resources and imposed unnecessary burdens on the opposing party. Unlike Mata or Hayes, the attorneys in this case quickly admitted the issue and cooperated, which the court acknowledged. Nonetheless, the submission of fabricated law—especially under counsel's signature—breaches core duties of candor and verification, warranting formal sanctions. The court warned that Utah’s judiciary cannot be expected to verify every citation and must be able to trust lawyers to do so

Sanction teardown · CA Utah, USA · 2025-05-22

Garner v. Kadince

What happened

In CA Utah, USA, a filing relied on ChatGPT to help draft legal argument. The court identified the following problems with the citations in that filing:

Fabricated (Case Law)

Fabricated appellate case cited in the Petition; opposing counsel showed it appears only in ChatGPT and not in any legal database, and the court sanctioned counsel for relying on fake precedent.

Which AI tool

ChatGPT. Note: Charlotin's public database records tool attribution only where a court order, brief, or reporting on the matter states it explicitly; "unidentified" or "implied" means the record indicates AI use but does not name a specific product — we do not guess.

Outcome

Not specified in source record. (monetary penalty: 1000 USD.)

Additional detail

AI UseThe fabricated citations originated from a ChatGPT query submitted by an unlicensed law clerk at Petitioner's law firm. Neither Counsel reviewed the petition’s contents before filing. The firm had no AI use policy in place at the time, though they implemented one after the order to show cause was issued.Hallucination DetailsChief among the hallucinations was Royer v. Nelson, which Respondents demonstrated existed only in ChatGPT’s output and in no official database. Other cited cases were also inapposite or unverifiable. Petitioner’s counsel admitted fault and stated they were unaware AI had been used during drafting.Ruling/SanctionThe court issued three targeted sanctions:Attorney fees: Respondents’ counsel are to submit an itemized bill; Counsel must pay within 10 days of receiptClient refund: Petitioner’s counsel must refund all fees paid by Mr. Garner in relation to the defective petitionCharitable payment: Counsel must donate $1,000 to “and Justice for all” within 14 days and file proof of payment with the courtKey Judicial ReasoningThe panel (Per Curiam) emphasized that the conduct, while not malicious, still diverted judicial resources and imposed unnecessary burdens on the opposing party. Unlike Mata or Hayes, the attorneys in this case quickly admitted the issue and cooperated, which the court acknowledged. Nonetheless, the submission of fabricated law—especially under counsel's signature—breaches core duties of candor and verification, warranting formal sanctions. The court warned that Utah’s judiciary cannot be expected to verify every citation and must be able to trust lawyers to do so

How Citation Safe would have caught this

Citation Safe runs three deterministic layers before a brief is filed: (1) does the citation exist against CourtListener's database of published opinions, (2) if quoted, does that exact language appear in the source, (3) does the cited case actually support the proposition it is cited for. Fabricated case citations fail Layer 1. Fabricated or misattributed quotations fail Layer 2 even when the underlying case is real. Misrepresented holdings — a real case cited for a proposition it does not support — are the target of Layer 3. None of these checks involve asking another language model whether the citation looks right; they are lookups and text-matches against the actual source, which is why a hallucinated citation has to survive a direct lookup against the authoritative source — not another model's opinion — to earn a VERIFIED stamp; our measured false-verify rate is published live at /quality.

Check a brief before you file it → · See our live false-verify rate

Source: https://www.damiencharlotin.com/documents/397/Garner_v._Kadince_UT_AC_USA_2025.pdf, via Damien Charlotin's public AI Hallucination Cases Database (CC0).