Sanction teardown · 9th Cir. CA, USA · 2024-03-22
Grant v. City of Long Beach
What happened
In 9th Cir. CA, USA, a filing relied on an unnamed/unconfirmed AI tool to help draft legal argument. The court identified the following problems with the citations in that filing:
-
Misrepresented (Case Law)Appellants claimed Hydrick examined a parent's false imprisonment claim over a child's unlawful removal; court notes Hydrick addresses SVP detainees' conditions of confinement and contains no 'parent'/'child' language.
-
Misrepresented (Case Law)Appellants claimed Wall addressed IIED against police for unlawfully removing a child; court explains Wall concerns a dentist's excessive force/false arrest/false imprisonment after an auto shop altercation, with no 'parent'/'child' language.
-
Fabricated (Case Law)Cited a non-existent case; court could not locate it.
-
Fabricated (Case Law)Cited a non-existent case; court could not locate it.
-
Misrepresented (Case Law)Court states appellants misrepresented the facts and holdings of Smith v. City of Salem.
-
Misrepresented (Case Law)Court states appellants misrepresented the facts and holdings of Yvonne L. v. N.M. Dep’t of Human Services.
-
Misrepresented (Case Law)Court states appellants misrepresented the facts and holdings of Smith v. City of Fontana.
-
Misrepresented (Case Law)Court states appellants misrepresented the facts and holdings of Wilkins v. City of Oakland.
-
Misrepresented (Case Law)Court states appellants misrepresented the facts and holdings of Harris v. Roderick.
-
Misrepresented (Case Law)Court states appellants misrepresented the facts and holdings of Mattos v. Agarano.
-
Misrepresented (Case Law)Court states appellants misrepresented the facts and holdings of Henderson v. City of Simi Valley.
-
Misrepresented (Case Law)Court states appellants misrepresented the facts and holdings of Johnson v. City of Seattle.
-
Misrepresented (Case Law)Court states appellants misrepresented the facts and holdings of Brooks v. City of Seattle.
-
Misrepresented (Case Law)Court states appellants misrepresented the facts and holdings of Devereaux v. Perez.
-
Misrepresented (Case Law)Court states appellants misrepresented the facts and holdings of Wallis v. Spencer.
Which AI tool
an unnamed/unconfirmed AI tool. 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
Striking of Brief + Dismissal of Appeal
Additional detail
AI UseThe appellants’ lawyer submitted an opening brief riddled with hallucinated cases and mischaracterizations. The court did not directly investigate the technological origin but cited the systematic errors as consistent with known AI-generated hallucination patterns.Hallucination DetailsTwo cited cases were completely nonexistent. Additionally, a dozen cited decisions were badly misrepresented, e.g., Hydrick v. Hunter and Wall v. County of Orange were cited for parent–child removal claims when they had nothing to do with such issues.Ruling/SanctionThe Ninth Circuit struck the appellants' opening brief under Circuit Rule 28–1 and dismissed the appeal. The panel emphasized that fabricated citations and grotesque misrepresentations violate Rule 28(a)(8)(A) requirements for arguments with coherent citation support.
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/236/Grant_v._City_of_Long_Beach_9th_C._CA_March_22_2024.pdf, via Damien Charlotin's public AI Hallucination Cases Database (CC0).