In June 2024, the Lipeles Law Group (“LLG”) filed a wage-and-hour class action in Los Angeles County against Harbor Distributing and related entities (the Ascensao action). Six months later, in December 2024, the same firm filed a nearly identical class action in San Francisco County against the same defendants on behalf of a different named plaintiff, Jaime Arguello Amaya Quinteros — an “overlapping, duplicate, copycat action,” in the words of the defense.

When Harbor moved to stay the San Francisco case under the doctrine of exclusive concurrent jurisdiction, LLG filed an opposition brief that proved to be a disaster. A contract attorney, James Sansone, had drafted it. Associate Jasmine Badawi reviewed it for flow but did not read the cited cases or check the citations; partners Kevin Lipeles and Thomas Schelly did not read the brief at all. The brief turned out to be riddled with fabrication: two case citations did not exist, and the trial court counted no fewer than eight fabricated quotations from five different cases, with the court observing that “[l]iterally every other purported quotation from a case in the brief is similarly fictitious.” The hallmarks pointed to undisclosed use of generative AI. Sansone denied using AI and insisted a Lexis citation check had validated everything — a claim the court found wholly incredible.

The trial court granted the stay and separately issued an order to show cause (“OSC”) re sanctions. After receiving declarations and holding a hearing — at which the firm’s lawyers apologized but could not coherently explain why they had filed a duplicative second action or opposed the stay, and continued to disclaim awareness of their own earlier-filed Los Angeles case — the court imposed sanctions under Code of Civil Procedure section 128.7(b). It ordered LLG and attorneys Lipeles, Schelly, and Badawi, jointly and severally, to pay $5,000 to Harbor and $1,000 to the court, to serve the order on the Los Angeles judge and on any San Francisco judge before whom they appeared for a year, and it referred the matter to the State Bar. The court called this “the worst example of misconduct by a lawyer” it had seen on the bench. Sansone, the court found, was “particularly blameworthy” for using AI and then denying it under oath, but the court held the firm bore “ultimate responsibility” because its name was on the brief.

The Court of Appeal affirmed the sanctions order in full in the published case of Quinteros v. Harbor Distributing, LLC (Lipeles et al., Objectors and Appellants), No. A174202 (June 2026). It held that LLG had forfeited two of its three arguments — the “safe harbor” procedural challenge and the objection to sanctions being paid to Harbor — and that the remaining argument, that the conduct did not warrant sanctions, had no merit because the trial court did not abuse its discretion.

The forfeited “safe harbor” challenge. LLG argued that section 128.7(c)(2) entitled it to 21 days to withdraw the offending opposition before sanctions could issue, and that the court’s compressed timeline denied that protection. The court rejected this for several reasons. First and foremost, LLG conceded it never raised section 128.7 in the trial court, and arguments not asserted below are forfeited. It declined to exercise its discretion to reach the forfeited issue, observing that a safe-harbor provision is not a matter of “vital public policy” (City of Rocklin v. Legacy Family Adventures-Rocklin, LLC, 86 Cal. App. 5th 713 (2022)). The court further invoked the invited-error doctrine: LLG had stipulated to the tentative ruling and never sought to cure, so any lost opportunity to withdraw was self-manufactured (Transport Insurance Co. v. TIG Insurance Co., 202 Cal. App. 4th 984 (2012)). Finally, the court noted LLG had buried its request for discretionary review in a footnote — itself a forfeiture (Golden Door Properties, LLC v. County of San Diego, 50 Cal. App. 5th 467 (2020)) — and raised the “important public policy” framing for the first time in its reply brief, too late to consider (American Indian Model Schools v. Oakland Unified School District, 227 Cal. App. 4th 258 (2014)).

The merits of the sanctions (abuse-of-discretion review). Reviewing under the deferential standard of Noland v. Land of the Free, L.P., 114 Cal. App. 5th 426 (2025), the court held the trial court acted well within its discretion. Section 128.7 lets a court sanction a filing made for an improper purpose or unwarranted by existing law; a claim is frivolous if any reasonable attorney would agree it is totally without merit (Kumar v. Ramsey, 71 Cal. App. 5th 1110 (2021)). The court surveyed the rapidly growing body of law on AI-fabricated citations, emphasizing that a fake opinion is not “existing law” and citing it abuses the adversary system — drawing on Noland, the seminal federal decision Mata v. Avianca, Inc., 678 F. Supp. 3d 443 (S.D.N.Y. 2023), and People v. Alvarez, 114 Cal. App. 5th 1115 (2025).

The forfeited objection to paying Harbor. LLG argued sanctions cannot be paid to an opposing party when the court issues an OSC on its own motion. The court found this forfeited too — not raised below (Doe WHBE 3 v. Uber Technologies, Inc., 102 Cal. App. 5th 1135 (2024)) and inadequately briefed, occupying only five lines with a miscited authority and no analysis (Dilbert v. Newsom, 101 Cal. App. 5th 317 (2024)). The court added that California Rules of Court, rule 2.30(b), independently permits awarding sanctions to an aggrieved party. It also observed in a footnote that the combined $6,000 sanction was far less than the “conservative” $10,000 imposed on a single attorney in Noland, and that LLG had improperly raised new authorities and a challenge to the amount for the first time at oral argument (Ramirez v. Charter Communications, Inc., 16 Cal. 5th 478 (2024)).