Sanction teardown · C.D. California, USA · 2025-05-06
Lacey v. State Farm General Insurance
What happened
In C.D. California, USA, a filing relied on CoCounsel, Westlaw Precision, Google Gemini to help draft legal argument. The court identified the following problems with the citations in that filing:
-
Fabricated (Case Law)Plaintiff cited a nonexistent case to support bad-faith discovery: Booth v. Allstate Ins. Co.; the Special Master could not locate it and it was later removed.
-
False Quotes (Case Law)Plaintiff attributed a fabricated quotation to National Steel Products; the Special Master found the quoted text did not exist in the opinion at the cited pin and the brief was revised.
-
Fabricated (Case Law)Plaintiff cited a nonexistent case: Davis v. City of Santa Ana; counsel later admitted it appears not to exist and should be removed.
-
False Quotes (Case Law)Plaintiff used a false quotation from Aetna v. Superior Court to assert a sword-and-shield principle; counsel admitted the quote does not appear in that case.
-
False Quotes (Case Law)Plaintiff misquoted Lipton v. Superior Court with a sword-and-shield line; counsel acknowledged the quotation was inaccurate.
-
False Quotes (Case Law)Plaintiff misquoted and mis-pincited Zurich American Ins. Co. v. Superior Court; counsel admitted the cited language and pin were inaccurate.
Which AI tool
CoCounsel, Westlaw Precision, Google Gemini. 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 briefs; denial of requested discovery relief; Large monetary sanctions jointly and severally against the two law firms (monetary penalty: 31100 USD.)
Additional detail
AI UseCounsel used CoCounsel, Westlaw’s AI tools, and Google Gemini to generate a legal outline for a discovery-related supplemental brief. The outline contained hallucinated citations and quotations, which were incorporated into the filed brief by colleagues at both Ellis George and K&L Gates. No one verified the content before filing. After the Special Master flagged two issues, counsel refiled a revised brief—but it still included six AI-generated hallucinations and did not disclose AI use until ordered to respond.Hallucination DetailsAt least two cases did not exist at all, including a fabricated quotation attributed to Booth v. Allstate Ins. Co., 198 Cal.App.3d 1357 (1989). Misquoted or fabricated quotes attributed to National Steel Products Co. v. Superior Court, 164 Cal.App.3d 476 (1985). Several additional misquotes and garbled citations across three submitted versions of the brief. Revised versions attempted to silently “fix” errors without disclosing their origin in AI output.Ruling/SanctionThe Special Master (Judge Wilner) struck all versions of Plaintiff’s supplemental brief, denied the requested discovery relief, and imposed:$26,100 in fees to reimburse Defendant for Special Master costs$5,000 in additional attorney’s fees to DefendantTotal monetary sanction: $31,100, payable jointly and severally by Ellis George LLP and K&L Gates LLPNo sanctions against individual attorneys due to candid admissions and remedial action, but strong warning issuedKey Judicial ReasoningThe submission and re-submission of AI-generated material without verification, especially after warning signs were raised, was deemed reckless and improper. The court emphasized that undisclosed AI use that results in fabricated law undermines judicial integrity. While individual attorneys were spared, the firms were sanctioned for systemic failure in verification and supervision. The Special Master underscored that the materials nearly made it into a judicial order, calling that prospect “scary” and demanding “strong deterrence.”
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/311/Lacey_v._State_Farm_General_Insurances_Co._D._Cal._May_6_2025.pdf, via Damien Charlotin's public AI Hallucination Cases Database (CC0).