New York courts are drawing a line on artificial intelligence in litigation: lawyers and litigants may use AI, but they remain responsible for every citation, quotation, legal proposition, and factual assertion filed with the court. The emerging trend in New York concerning the use of AI in litigation is no longer theoretical. Courts are warning litigants, striking papers, imposing monetary sanctions, and, in some cases, considering referral to the appropriate attorney grievance committees.
The central problem is not simply that a lawyer or party used artificial intelligence. The problem is the filing of AI-generated hallucinations, misgrounded authority, fake cases, inaccurate quotations, and legal propositions that the cited cases do not support.
The Trend: Verify AI or Risk Sanctions
The recent New York cases show a rapidly developing judicial response to AI-generated legal research errors. In Deutsche Bank Natl. Trust Co. v LeTennier, 2026 NY Slip Op 00040 (3rd Dept 2026), the Appellate Division, Third Department imposed sanctions after appellate filings contained fabricated cases and misrepresented authority. The court made the practical rule clear: generative AI is not prohibited, but every citation must be verified before submission.
The Second Department tackled AI misuse in Matter of Julien v Arthur, 2026 NY Slip Op 03308 (2d Dept 2026), the Appellate Division, Second Department imposed a monetary sanction against a pro se litigant who submitted a brief containing fictitious legal authority. The sanction was modest, but the message was important: even one fabricated case citation can constitute frivolous conduct.
“The issue has also surfaced at oral argument, where courts have questioned whether opposing counsel should alert the court when an adversary’s papers contain hallucinated authority.” [Editor’s Note: I omitted the case name and oral argument video to avoid adding to the list of posts and comments about the attorneys on the appeal]
At the trial-court level, New York Supreme Court decisions have addressed different aspects of AI misuse. The recent New York AI decisions should not be read as isolated citation-error cases. Together, they show an emerging judicial progression: courts are first identifying AI-generated hallucinations, then warning litigants, striking defective submissions, imposing sanctions, and, in the most serious cases, considering whether counsel’s conduct should be referred for professional discipline.
That progression begins with pro se filings, but it does not end there. In Keefe v BMakin Film Ltd., 2026 NY Slip Op 31794(U) (Sup Ct, Kings Cty 2026), the court addressed papers containing AI-generated hallucinations and misgrounded authority. Although the plaintiff appeared without counsel, the court did not treat that status as a license to file false legal authority. The conduct was deemed frivolous under 22 NYCRR 130-1.1, but the court stopped short of imposing a monetary sanction, issuing a warning instead. The point was still unmistakable: pro se litigants may use AI, but they remain responsible for what they file.
Michaelis v Wilmington Savings Fund Society, FSB, 2026 NY Slip Op 26048 (Sup Ct, Dutchess Cty 2026), adds a procedural dimension to that rule. There, the problem was not merely that a pro se litigant used AI in an affirmation. The problem was that the filing contained hallucinated cases and misrepresentations in a court that had Part Rules addressing AI use. For New York litigators, Michaelis is a reminder that AI compliance is no longer limited to general ethical duties. Counsel must also check the assigned judge’s individual rules before submitting AI-assisted filings, affirmations, or memoranda of law.
Other courts have focused less on the litigant’s status and more on the burden imposed on the court system. In Abello v NYP Holdings, Inc., 2026 NY Slip Op 31021(U) (Sup Ct, NY Cty 2026), the court confronted a memorandum of law filled with AI-generated hallucinations. It declined to impose sanctions, but its warning was pointed. When AI-generated research forces a judge or adversary to track down nonexistent cases or inaccurate propositions, it wastes limited judicial resources and undermines the integrity of motion practice.
The remedy became more concrete in Gully v Varghese, 2026 NY Slip Op 50864(U) (Sup Ct, Albany Cty 2026). There, the court found that fake case citations appeared to be AI hallucinations, struck the offending papers, denied summary judgment, and warned that future violations could lead to more serious consequences. Gully is important because it shows that AI misuse can have immediate litigation consequences. A party does not merely risk embarrassment or a judicial reprimand; defective AI-generated submissions can cost the party the motion.
The Grievance Committee Risk
The most serious development is the growing willingness of courts to consider professional discipline.
In Kleyman Law Group, P.C. v Kaloidis, 2026 NY Slip Op 31557(U) (Sup Ct, Kings Cty 2026), the court described submissions containing numerous fictitious cases and misrepresented authorities. What makes the decision especially important for New York litigators is that the court discussed whether counsel’s conduct should be referred to the appropriate attorney grievance committee.
That is the practical danger for attorneys. New York courts are not treating AI errors as harmless citation mistakes when they are filed in court papers. They implicate candor to the tribunal, competence, supervision, diligence, and the duty to correct false legal submissions.
For law firms, the lesson is straightforward: an AI policy is no longer optional. Lawyers need a verification process for every AI-assisted filing, especially motion papers, appellate briefs, affirmations, and memoranda of law.
The New Issue: Discovery of AI Prompts and Outputs
The newest development is not just sanctions. It is discovery.
In Assini v Hayward, 2026 NY Slip Op 26086 (Sup Ct, Nassau Cty 2026), the court addressed a subpoena directed to OpenAI seeking prompts, inputs, uploaded materials, generated outputs, and AI communications allegedly used to draft or revise filings, motions, sworn statements, and litigation communications.
The court quashed the subpoena. It recognized that AI prompts and outputs used by a litigant in preparing legal papers may raise work-product, litigation-preparation, privacy, and confidentiality concerns. The decision is important because it shows that AI use in litigation may create a new discovery battleground: whether prompts, outputs, uploaded documents, and AI-generated drafts are discoverable.
But the court’s protection was not a free pass and came with a warning. The court still noted that the litigant’s AI use had not gone unnoticed, referred him to 22 NYCRR Part 161, and warned that failure to comply with the court rules governing artificial intelligence may result in sanctions.
The Impact on New York Litigators and Claims Professionals
The developing New York AI caselaw points in one direction: courts are not banning artificial intelligence, but they are demanding accountability.
For attorneys, the risk is professional. AI hallucinations can lead to sanctions, struck papers, fee awards, public criticism, and potential grievance referral.
For claims professionals and insurers, the risk is strategic. AI-generated briefing errors can weaken a motion, delay resolution, increase litigation cost, and create avoidable exposure. A strong defense position can be damaged by a brief that cites fake cases or misstates real ones.
For pro se litigants, the message is equally clear: courts will not excuse fabricated authority simply because the party relied on an AI tool.
Practical Considerations
New York lawyers using artificial intelligence should verify every case citation through an authoritative legal database, read every cited decision, confirm every quotation, and ensure that each case actually supports the proposition for which it is cited.
Law firms should also consider a written AI-use protocol requiring attorney review, citation checking, quote verification, prompt confidentiality safeguards, and immediate correction if an AI-generated error is discovered.
Counsel should also carefully review the case law opposing counsel relies on. If there are hallucinations or inaccuracies, you should notify the court of them.
The New York trend is now unmistakable: artificial intelligence can assist legal research and drafting, but it cannot replace professional judgment. In New York courts, the lawyer who files the paper owns the citation, the quotation, and the representation made to the court.
Related Posts
New York Appellate Division Sanctions AI-Generated Fake Case Citations
New York Sanctions Attorney for AI-Generated Fake Case