Sanction teardown · N.D. New York, USA · 2025-07-09
Gurpreet Kaur v. Captain Joel Desso
What happened
In N.D. New York, USA, a filing relied on Claude Sonnet 4 to help draft legal argument. The court identified the following problems with the citations in that filing:
-
False Quotes (Case Law)Brief claimed Thuraissigiam included a sentence limiting its holding and the quoted line; court confirmed the quote is nowhere in the decision.
-
False Quotes (Case Law)Brief quoted Landon as stating a due-process proposition; court stated the quotation does not exist in the case.
-
False Quotes (Case Law)Brief asserted Reno contains a phrase about jurisdiction-stripping not barring challenges; court found the language is nowhere in Reno.
-
False Quotes (Case Law)Brief cited Wong Wing for a broad constitutional proposition; court found the quoted language does not exist in Wong Wing.
Which AI tool
Claude Sonnet 4. 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
Monetary and professional sanctions (monetary penalty: 1000 USD.)
Additional detail
Counsel confessed having Claude Sonnet 4 to draft a legal submission, which included fabricated quotations from legal authorities. Counsel said he was pressed by time.After holding that there is "no reason to distinguish between the submission of fabricated cases and the submission of fabricated quotations from real cases. In both postures, the attorney seeks to persuade the Court using legal authority that does not exist", the court held that Mr. Desmarais had violated Rule 11 of the Federal Rules of Civil Procedure by failing to verify the accuracy of the AI-generated content. Counsel was found to have acted in subjective bad faith, as he was aware of the potential for AI to hallucinate legal citations and failed to take corrective action even after the government pointed out the errors.The court imposed a $1,000 monetary sanction and required Counsel to complete a CLE course on the ethical use of AI in legal practice and notify his client of the issue.
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/581/Kaur_v._Desso_NDNY_9_July_2025.pdf, via Damien Charlotin's public AI Hallucination Cases Database (CC0).