Background
The underlying personal injury action arose from a trip-and-fall on a loose brick in a tree well on a sidewalk abutting a Whitman Owner Corp. property in Brooklyn. Plaintiff Judith Landberg sued both the property owner and the City of New York. Supreme Court, Kings County granted both defendants’ summary judgment motions in December 2024, and on June 3, 2026 the Second Department affirmed.
That affirmance prompted a more consequential proceeding. During oral argument in May 2026, the appellate panel confronted plaintiff’s attorney, Michael Sanders of the Law Offices of Michael S. Lamonsoff, PLLC, about citations in the appellant’s brief that appeared nowhere in any law reporter. When the court raised concerns about fabricated authority, Sanders — without taking the offered 15-minute recess to investigate — asserted that his citations came from Westlaw, Lexis, a book, or a prior brief. On the same day, the court issued an order to show cause calling on Sanders and his firm to explain why sanctions should not be imposed under 22 NYCRR 130-1.1 for filing briefs containing fabricated and/or erroneous legal authority.
In a subsequent affirmation, Sanders admitted what he had denied at oral argument: the fictitious citations originated during “AI-assisted research” using “one of the free ones available to the public.” He acknowledged that he “negligently failed to verify” the AI output before filing, and that when the panel questioned him he “should have stated that they were hallucinated by artificial intelligence. I was genuinely scared at the time.”
The Court’s Holding
The court catalogued four categories of error in Sanders’s brief. First, the brief twice cited the wholly nonexistent case of Xiang Fu Ji v City of New York (13 NY3d 791) for the proposition that Administrative Code § 7-210 — which assigns sidewalk-maintenance liability to abutting property owners — is a remedial statute to be “construed liberally.” In reality, the Court of Appeals has held the opposite: because § 7-210 is “in derogation of common law,” it “must be strictly construed.” Second, the brief fabricated a second nonexistent case, Hausser v Giunta (88 AD3d 969), for a different proposition about § 7-210. Third, the brief cited the fictitious Lack v Lack (841 NYS2d 403) for the de novo standard of review on summary judgment. Fourth, the brief misrepresented what two real cases actually decided.
Assessing sanctions, the court drew on 22 NYCRR 130-1.1, which authorizes financial sanctions against attorneys who engage in “frivolous conduct,” including asserting “material factual statements that are false.” The court distinguished Matter of Julien v Arthur (2026 NY Slip Op 03308) — in which a pro se litigant received a $250 sanction for citing a single nonexistent case — and stressed the greater culpability of a licensed attorney with access to Westlaw and Lexis. The court also noted that Sanders’s lack of candor at oral argument “demonstrated his professional immaturity, arrogance, and profound lack of respect for the judicial system.” Separately, it credited the law firm’s general counsel’s statement that Sanders acted contrary to firm policy and that the firm had scheduled mandatory AI-ethics training for all staff. The firm’s sanction was set lower accordingly.
The sanctions ordered: $8,000 against Sanders personally, and $2,500 against Law Offices of Michael S. Lamonsoff, PLLC — both payable to the Lawyers’ Fund for Client Protection within 20 days. The court reserved the question of whether disciplinary referral to the Attorney Grievance Committee was also warranted.
Key Takeaways
- Using a generative AI tool for “supplemental research” and filing unverified output is sanctionable under 22 NYCRR 130-1.1 — the rule reaches fictitious citations regardless of how they were generated.
- The court quoted the Third Department’s recent holding in Deutsche Bank Natl. Trust Co. v LeTennier that GenAI use “in no way abrogates an attorney’s obligation to fact check and cite check every document filed with a court” — making that principle now authoritative in the Second Department as well.
- Denying AI use to the court when confronted, then later admitting it in a written affirmation, compounded the sanction and may separately implicate Rule 3.3(a)(1) (candor toward the tribunal).
- A law firm can face vicarious sanctions even when its employee violated firm policy, if its name appeared on the offending brief — though good-faith remedial steps (training, policy enforcement) will reduce the amount.
Why It Matters
This decision joins a rapidly growing body of New York authority — Matter of Julien v Arthur, Grymes Dev. Co. v Fodera, and now Landberg — establishing that AI-generated citation errors carry real financial and professional consequences in this state. The stakes are highest where the AI hallucination does not merely invent a case but inverts actual doctrine: here, the fabricated quotation told the court to read a statute “liberally” that the Court of Appeals had expressly required to be read “strictly.” That kind of invisible error — plausible-looking, confidently stated, and wrong in a decisive way — is precisely what makes unverified GenAI output dangerous in an adversarial proceeding.
For New York litigators: every citation in every brief needs human verification against a primary source, not merely a plausible AI summary. The court expressly noted that stories about AI unreliability “have been widespread in the media for years, putting all attorneys on notice” — meaning ignorance of the risk is no longer a mitigating factor. Firms should audit their AI-use policies now, before a fabricated citation surfaces in an already-lost appeal and the only question remaining is how large the sanction will be.