The Ninth Circuit fines two lawyers $2,500 each and suspends them for six months over AI-hallucinated briefs, but the sanction is for candor, not for using AI

The problem

An immigration petition for review reached the Ninth Circuit (Lnu, et al. v. Blanche, No. 24-4790), where petitioners sought review of a Board of Immigration Appeals order dismissing their asylum, withholding-of-removal, and Convention Against Torture claims (source). Petitioners’ counsel Mike Singh Sethi filed an opening brief with multiple fabricated citations and quotations: he “cited two cases that do not exist and never existed,” namely “Eduardo v. Garland, 28 F.4th 742 (9th Cir. 2022)” and “Lay v. Holder, 729 F.3d 962 (9th Cir. 2013),” and “twice attributed quotations to real opinions in which the quoted language does not appear” (source). When the fabrications surfaced, the attorneys called them typographical errors and repeatedly denied that generative AI had produced them (source).

What was built

No system was built or deployed here. This is an AI-governance honest-negative: a federal appeals court’s response to unverified generative-AI output submitted as law. The panel (Judges Richard A. Paez, Carlos T. Bea, and Danielle J. Forrest; Order “For Publication,” filed June 3, 2026) found the fabrications and inaccuracies “were the result of hallucinations by generative AI,” but was careful to say it was not sanctioning the mere use of AI (source). “The rules are not violated at the point of research and drafting, but at the point of signing and filing,” the panel wrote (source). The discipline turned on conduct after the filing: the attorneys “knowingly or recklessly made false statements to this Court” by characterizing the nonexistent cases as typographical slips (source). The order names no specific AI product (source).

The outcome

The monetary sanction: $2,500 on each attorney, $5,000 combined. Acting “under Federal Rule of Appellate Procedure 46(b) and Ninth Circuit Rule 46-2,” the panel ordered that “Sethi and Rounds are each personally sanctioned in the amount of $2,500,” payable to “Clerk, US Courts” within 21 days (source, source).

The suspension: six months each. The order states that “Sethi and Rounds are hereby suspended from practice before this Court for a period of six months starting ten days after this Order is filed” (source).

The ongoing conditions. For all future filings, the panel ordered that “Sethi, Rounds, and all attorneys at the Firm” include a statement under penalty of perjury addressing whether generative AI was used, disclosing the name of the tool, and certifying that the signing attorney has personally reviewed the filing and that every citation and quotation refers to existing authority (source). The panel also referred the matter to the State Bar of California (source).

Why the discipline was this heavy. The panel was explicit that the suspension was not a penalty for using AI: “the gravity of discipline we impose, including the temporary suspension of practice, is owed to this repeated failure of candor” (source).

How this was verified

This case carries a green badge under TIN’s court-adjudicated standard: the origin is a published, self-authenticating federal appellate order, and no client-confirmation call is booked against the court. Every figure was re-checked live on 2026-08-16. The $2,500-each sanction and the six-month suspension are corroborated by two mutually independent newsrooms, Bloomberg Government and the Metropolitan News-Enterprise, and Reason/The Volokh Conspiracy reproduces the order text, so the fabrications, the panel, the payment terms, and the certification order all trace to the primary and to independent narration.

One figure was corrected on re-fetch. An earlier draft of this file described the AI-disclosure certification as lasting “two years.” The order’s own language, as reproduced by Volokh, imposes the certification on “all future filings” with no stated end date, so the “two years” figure has been removed rather than carried forward. The court’s finding that the fabrications “were the result of hallucinations by generative AI” is quoted from the published order; no specific AI product is named anywhere in the record, and no claim here asserts one.

Related case files

The same appellate pattern, a federal circuit court finding AI-hallucinated citations and reaching for discipline rather than a fine, appears in Akerlund v. Atlas Air, where the Eleventh Circuit referred counsel to its Committee on Lawyer Qualifications and Conduct after briefs “replete with fake and hallucinated citations.” California is the connecting thread to Noland v. Land of the Free, the state’s first published appellate opinion on AI-fabricated citations, where a $10,000 sanction followed the same failure to verify. And Gauthier v. Goodyear shows the earlier, lighter end of the same escalation, a $2,000 sanction and mandatory AI education, before courts began adding suspensions and standing certification orders.

Sources

Cited in this case file. Tier 1 = the primary court order; Tier 2 = independent press naming the parties; Tier 3 = a practitioner explainer reproducing order language. Each figure was checked against the live source on 2026-08-16.

1. U.S. Court of Appeals for the Ninth Circuit, “Lnu, et al. v. Blanche, No. 24-4790” (published Order), 2026-06-03 (Tier 1, primary; the adjudicating court stating its own sanction). https://cdn.ca9.uscourts.gov/datastore/opinions/2026/06/03/24-4790.pdf · archived
2. Reason / The Volokh Conspiracy, “Ninth Circuit on AI Hallucinations,” 2026-06-03 (Tier 2, independent; reproduces the order text and quotes the panel). https://reason.com/volokh/2026/06/03/ninth-circuit-on-ai-hallucinations/
3. Bloomberg Government, “Ninth Circuit Warns of AI Hallucinated Briefs in Sanctions Order,” 2026-06 (Tier 2, independent press naming the parties and figures). https://news.bgov.com/legal-ops-and-tech/ninth-circuit-warns-of-ai-hallucinated-briefs-in-sanctions-order
4. Metropolitan News-Enterprise, “Two O.C. Attorneys Suspended Relating to AI Fabricated Cites,” 2026-06-04 (Tier 2, independent). http://www.metnews.com/articles/2026/sanctions_060426.htm
5. ComplianceHub, “The Point of Signing: the Ninth Circuit’s Precedential AI-Hallucination Sanctions,” 2026 (Tier 3, practitioner explainer reproducing the order’s “signing and filing” language and the State Bar referral). https://compliancehub.wiki/ninth-circuit-ai-hallucination-sanctions-precedent-2026/