Quinteros v. Harbor Distributing, LLC

Facts

In June 2024, the Lipeles Law Group (LLG) filed a putative wage-and-hour class action against Harbor Distributing and affiliated entities in Los Angeles (the Ascensao action), later adding a PAGA claim. (slip opn. at pp. 2–3.) In December 2024, the same firm filed a “substantially duplicative” class action against the same defendants in San Francisco (the Quinteros action). (slip opn. at pp. 2–3.) Harbor moved to stay the later San Francisco action under the doctrine of exclusive concurrent jurisdiction, attaching the Los Angeles complaint. (slip opn. at pp. 3–4.)

LLG’s opposition — drafted by contract attorney James Sansone and signed by associate Jasmine Badawi, with partners Kevin Lipeles and Thomas Schelly also named — argued that “California law … requires complete—not partial—overlap.” (slip opn. at pp. 3–4.) The trial court found the brief “contain[ed] non-existent citations, fabricated quotations, and seriously misrepresent[ed] controlling authority”: two cited cases did not exist, and the brief contained “no fewer than eight fabricated quotations.” (slip opn. at pp. 4–5.) The court observed that fabricated quotations created by a tool “such as ChatGPT … are even more insidious than quotations from ‘hallucinated’ cases because they are attributed, falsely, to actual reported (if miscited) cases.” (slip opn. at p. 5.) At the hearing, partners admitted they had not read the brief and the associate admitted she “did not fact check or cite check” it; the court called it “the worst example of misconduct by a lawyer that I … [have] ever seen.” (slip opn. at pp. 7–9.) Sansone, in a declaration, denied using generative AI and claimed a Lexis citation check validated the authorities. (slip opn. at pp. 6–7.)

Procedural history

The court issued an order to show cause and, on July 16, 2025, sanctioned LLG and the three named attorneys, jointly and severally, $5,000 payable to Harbor and $1,000 to the court, under section 128.7, subdivision (b), and counsel’s ethical obligations, also ordering notice to other courts. (slip opn. at pp. 11–13.) LLG appealed, arguing the court violated section 128.7’s safe-harbor provision, that the conduct did not merit sanctions, and that sanctions payable to Harbor were improper. (slip opn. at p. 13.) The Court of Appeal affirmed. (slip opn. at p. 26.)

Issue

Did the trial court abuse its discretion in sanctioning a firm and its attorneys of record under section 128.7 for an opposition brief containing fabricated, AI-generated citations and quotations prepared by an outside contract attorney — and did the firm preserve its objections that the court failed to honor the statute’s 21-day “safe harbor” and improperly directed sanctions to the opposing party? (slip opn. at pp. 13, 19.)

Holding

Reasoning

1. Forfeiture of the safe-harbor argument. The court applied the bedrock rule that “‘arguments not asserted below are waived,’” and noted that “multiple courts have deemed arguments based on the failure to comply with section 128.7’s safe harbor provision forfeited when not raised before the trial court.” (slip opn. at pp. 14–15, citing In re Marriage of Falcone & Fyke (2008) 164 Cal.App.4th 814, and Peake v. Underwood (2014) 227 Cal.App.4th 428, 447.) The safe harbor is not “a matter of vital public policy,” so no exception applied. (slip opn. at pp. 15–16, citing City of Rocklin v. Legacy Family Adventures-Rocklin, LLC (2022) 86 Cal.App.5th 713, 735.) The court distinguished Barnes v. Department of Corrections (1999) 74 Cal.App.4th 126 — a post-judgment, party-filed sanctions motion under subdivision (c)(1) — from this court-initiated OSC under subdivision (c)(2) “at the comparative start of the case.” (slip opn. at pp. 15–17.)

2. Any cure opportunity was lost by LLG’s own conduct. The court emphasized that LLG could have investigated the Los Angeles action and withdrawn its opposition before the reply, before the tentative ruling (which expressly flagged a forthcoming sanctions OSC), or in response to the OSC — but did not, and instead “stipulated to the tentative ruling.” (slip opn. at pp. 16–17.) Under the invited-error doctrine, a party “‘by its own conduct’” that “‘induces the commission of error’” cannot complain of it on appeal. (slip opn. at pp. 16–17, quoting Transport Ins. Co. v. TIG Ins. Co. (2012) 202 Cal.App.4th 984, 1000.) The court also faulted LLG for relegating its request for discretionary review to a footnote and raising new arguments only in reply. (slip opn. at pp. 17–18.)

3. The merits: sanctions were warranted. Reviewing for abuse of discretion (Noland, supra, 114 Cal.App.5th at p. 438), the court joined the “ever-growing body of law” holding that “filing briefs containing fabricated legal authority is sanctionable.” (slip opn. at pp. 20–21.) A “fake opinion is not ‘existing law,’” and reliance on fabricated authority “is an abuse of the adversary system.” (slip opn. at p. 20, quoting Noland and Mata v. Avianca, Inc. (S.D.N.Y. 2023) 678 F.Supp.3d 443.) The duty of competence “requires that lawyers review and evaluate the output produced by generative AI before submitting it to a court.” (slip opn. at p. 22, citing ABA Formal Opn. No. 2024-512 and the State Bar’s “Practical Guidance for the Use of Generative Artificial Intelligence in the Practice of Law.”)

4. Counsel of record cannot outsource responsibility. The decisive point was non-delegation. Although the contract attorney drafted the brief and apparently used AI while denying it, LLG “is ‘not absolved of responsibility merely because [it] outsourced preparation of the brief.’” (slip opn. at p. 22.) “Counsel of record’s names appeared on the brief, and one of them signed it,” so by section 128.7, subdivision (b)(2) they certified that “after an inquiry reasonable under the circumstances” the contentions were “warranted by existing law” — a certification they violated by failing to read the brief at all (the partners) or to “substantively review or cite check” it (the associate). (slip opn. at pp. 22–23.)

5. The remaining challenge and the amount. LLG forfeited its objection to the award payable to Harbor by not raising it below and by inadequate briefing; in any event, California Rules of Court, rule 2.30(b) permits a sanction “to the court or an aggrieved [party], or both.” (slip opn. at pp. 23–26.) The court noted the $6,000 joint-and-several sanction was “far less than the ‘conservative’ $10,000 sanction issued against a single attorney in Noland,” and saw “no error in the amount.” (slip opn. at p. 24.)

Significance

Quinteros is a leading California appellate statement on attorney responsibility for generative-AI “hallucinations,” and it lands with particular force on the employment bar, where contract attorneys and AI drafting tools are increasingly common. Its core holdings are bright lines: every citation in a filed paper must be personally read and verified by the attorney submitting it; that duty “cannot be delegated to any form of technology”; and counsel of record bear ultimate responsibility even when the work was outsourced to another lawyer who used — and concealed — AI. The decision also reinforces ordinary preservation discipline: a litigant who wants the protection of section 128.7’s safe harbor, or who objects to the form of a sanction, must say so in the trial court. For any firm using AI in research or drafting, Quinteros is a direct warning that a signature on a brief is a certification of its accuracy, and that no software citation-checker, and no junior or contract attorney, can substitute for the signer’s own verification.

Key quotes

“[N]o brief, pleading, motion, or any other paper filed in any court should contain any citations—whether provided by generative AI or any other source—that the attorney responsible for submitting the pleading has not personally read and verified.” (Quinteros, supra, No. A174202 (slip opn. at p. 20), quoting Noland, 114 Cal.App.5th at p. 431.)

“[A]ttorneys cannot delegate this responsibility to any form of technology; this is the responsibility of a competent attorney.” (Quinteros, supra, No. A174202 (slip opn. at p. 21), quoting Alvarez, 114 Cal.App.5th at p. 1119.)

Read the full opinion (official slip opinion) →

Practice pointer

Open questions

Because it resolved the safe-harbor issue on forfeiture, the court did not decide whether a court-initiated OSC under section 128.7, subdivision (c)(2) must always afford a full 21-day withdrawal period before sanctions, or how that period operates when the offending paper concerns a motion already resolved — a question on which the parties disputed Barnes’s reach. (slip opn. at pp. 13–17.) The opinion also leaves open the precise allocation of responsibility (and any direct sanction) as between counsel of record and an outside contract attorney who uses and conceals AI; here the trial court found the contract attorney “particularly blameworthy,” but the appeal concerned only the sanctions against the firm. (slip opn. at pp. 12, 22.) Finally, the decision applies, but does not exhaustively define, the “duty to review and evaluate” AI output drawn from the ABA and State Bar guidance. (slip opn. at p. 22.)