Ninth Circuit Warns of AI Hallucinated Briefs in Sanctions Order

The country’s largest federal appeals court sanctioned and suspended two attorneys who failed to disclose inaccuracies in their legal briefs came from generative AI hallucinations.

The US Court of Appeals for the Ninth Circuit said Wednesday the disciplinary order is a warning to members of the court’s bar to be cautious of overreliance on generative AI when drafting legal briefs.

“Lawyers using generative AI must thus be aware of the tendency of generative AI to make these mistakes and guard against them,” the ruling said.

The order imposes sanctions of $2,500 each against attorneys Mike Sethi and William Rounds of Orange County, Calif., who submitted briefs in an immigration case containing citations to opinions that didn’t exist and fabricated quotes from existing cases.

The attorneys are also suspended from practicing before the Ninth Circuit for six months and they must send a copy of the order to their clients, opposing counsel, and presiding judge in all of their other cases.

The attorneys initially told the court the mistakes were simple typographical errors before conceding they were likely the result of the brief writer, an unlicensed law school graduate at the firm, using unauthorized AI.

Legal research companies and generative AI labs have been racing to roll out AI tools for the legal industry, with both Anthropic PBC and OpenAI pushing in recent months to add legal plugins for their flagship chatbots. The new technology has drawn a nationwide increase in sanctions against lawyers filing briefs with AI-related hallucinations and inaccuracies.

In February, the Fifth and Tenth circuits imposed similar sanctions against attorneys for using AI tools to draft briefs containing inaccuracies.

The three-judge panel behind the ruling said the Ninth Circuit, which covers much of the American West, has previously stricken a brief that was almost entirely fabricated in another case. But it hadn’t yet explained the proper consequences when parts of a brief include hallucinations or mistakes.

“We first clarify what this order is not about,” the court said. “We do not sanction Sethi and Rounds for the simple fact that they or their subordinates used generative AI.”

The ruling said the Federal Rules of Appellate Procedure and the Ninth Circuit rules don’t directly regulate the process by which attorneys produce a legal filing, but they must still follow procedural and ethical rules.

AI-generated fabrications are the most notorious kind of hallucination, “but inaccuracies may prove more dangerous to our profession in the long run” because they are likely to go unnoticed, the court said.

The opinion cited a law article finding that AI tools from Westlaw and Lexis hallucinated 17% and 33% of answers, respectively, to a representative set of questions run in 2024.

Judges Richard Paez, Carlos Bea, and Danielle Forrest sat on the panel.

The case is Lnu v. Blanche, 9th Cir., No. 24-4790, 6/3/26.