Sanctions and bar referrals show real consequences for AI

By now, most attorneys are familiar with the limitations of artificial intelligence--particularly its tendency to produce fabricated citations and unreliable authority--but the California Court of Appeal has now made clear that those "hallucinations" can carry very real consequences.

In Noland v. Land of the Free, L.P. (B331918, Sept. 12, 2025), Division Three of the Second District sanctioned the plaintiff/appellant's attorney $10,000 and reported him to the state bar after finding that nearly all of the legal quotations in his appellate briefs were invented by generative AI tools and had never appeared in any published case.

And the Court was clear as to the reasons why it was publishing the opinion: "We ... publish this opinion as a warning. Simply stated, no 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." (emphasis in original).

Amir Mostafavi, who ironically boasts of his extensive experience in legal malpractice matters on his website, used generative AI tools to draft both the appellate opening brief and the reply brief. This included ChatGPT, Claude, Gemini and Grok. The use of these AI tools caused "nearly all of the quotations in plaintiff's opening brief, and many of the quotations in plaintiff's reply brief, [to] have been fabricated." In short, "the quotes plaintiff attributes to published cases do not appear in those cases or anywhere else. Further, many of the cases plaintiff cites do not discuss the topics for which they are cited, and a few of the cases do not exist at all. These fabricated legal authorities were created by generative artificial intelligence (AI) tools that plaintiff's counsel used to draft his appellate briefs."

The underlying case was a simple employment lawsuit involving wage-and-hour and harassment claims. After a tortured history, the trial court granted summary judgment primarily on the basis that the plaintiff was an independent contractor rather than an employee.

The plaintiff filed a notice of appeal from the order granting summary judgment, which is not an appealable order. As explained in footnote 3, the Court of Appeal directed the appellant's counsel to provide an appealable judgment. Perhaps forecasting the level of dilatory behavior of Mostafavi, "[a]lthough more than 18 months have passed since this court advised plaintiff of the need to obtain a judgment, plaintiff's counsel has not obtained one." 

Even though the court explicitly stated it was within its discretion to dismiss the appeal for lack of standing, it decided not to do so and instead construed the order as a judgment. The Court of Appeal curtly addressed the underlying appeal, finding that the trial court appropriately granted summary judgment.

The Court of Appeal then went through meticulous and painful detail as to all of the cases cited in the plaintiff's briefing. In total, of the 23 case quotations in plaintiff's opening brief, 21 were fabricated. This was in addition to the brief being "peppered with inaccurate citations that do not support the propositions for which they are cited."   The use of hallucinations and inaccuracies in the reply was similar.

In response to the Court's Order to Show Cause, Mostafavi admitted to using AI, but "asserted that he had not been aware that generative AI frequently fabricates or hallucinates legal sources and, thus, he did not 'manually verify the quotations against more reliable sources.'" While he "accepted responsibility" for the fabrications, Mostafavi also doubled down, asserting that "the majority of citations are accurate and support the propositions that were being advanced" and that the appeal "stands on meritorious arguments that are fully supported by the record." 

The Court of Appeal hammered Mostafavi. The Court of Appeal explained: "To state the obvious, it is a fundamental duty of attorneys to read the legal authorities they cite in appellate briefs or any other court filings to determine that the authorities stand for the propositions for which they are cited." (emphasis in original). The Court went on to explain that "although there is nothing inherently wrong with an attorney appropriately using AI in a law practice--before filing any court document, an attorney must carefully check every case citation, fact, and argument to make sure that they are correct and proper." 

The Court of Appeal compared using AI to supervising a junior attorney, law clerk or intern. The supervising attorney must still carefully check the veracity of all case citations before filing.

Accordingly, the appellate court found that the appeal was frivolous as it rested on erroneous legal foundation. It also clearly violated the Rules of Court. As such, the Court of Appeal imposed a "conservative sanction" of $10,000. The court also directed both Mostafavi and the court clerk to separately serve a copy of the opinion to the state bar. Sanctions were not awarded to opposing counsel because, somewhat remarkably, the defendant's counsel filed its respondent's brief without noticing that the plaintiffs' briefing had used such fabricated authority.

The Court of Appeal's published opinion stands as a public warning: if you let a chatbot write your brief, you may soon find yourself writing sanction checks--and answering to the State Bar--instead of persuading a court.