AI Regulation Tracker  /  For Attorneys

NY Part 161: The New AI Verification Rule for Litigators

New York now has a binding statewide court rule on AI in filings, and the point litigators keep missing is that it demands verification, not disclosure.

By Anthony Guerriero  ·  Reviewed by The Leveraged Years Editorial Desk  ·  Published July 1, 2026  ·  Last updated July 1, 2026

New York's Part 161, effective June 1, 2026, is a statewide court rule that permits attorneys and parties to use AI when preparing filings but does not require disclosing that use. Instead, it requires the signing attorney to carefully review the paper and independently confirm it contains no fabricated or fictitious cases, statutes, or other material. The rule binds every New York State courtroom, and individual judges may add stricter AI requirements in their own part rules, so obligations can vary by courtroom.

On June 1, 2026, the New York State Unified Court System put Part 161 into effect, the state's first statewide rule addressing artificial intelligence in court filings. If you write briefs in New York, this is now the backdrop for every submission you sign. The rule was adopted on March 25, 2026, and it settles a question that had been hanging over litigators since the first wave of hallucinated-citation sanctions: is using AI to draft a filing allowed, and if so, on what terms?

Here is the part that trips people up. The locked framing everyone is using calls this a disclosure rule. Read the text and you find the opposite. Part 161 permits AI use and declines to require that you disclose it. What it requires instead is verification, and that distinction is the whole story for how you should run your desk starting now.

What Part 161 actually says

Part 161 broadly permits attorneys and parties to use AI when preparing briefs, affidavits, pleadings, and other documents for submission to a New York court. The policy position, as the New York State Bar Association describes it, is that AI use should not be prohibited so long as it stays consistent with a lawyer's existing professional duties.

The operative obligation is a duty to review. Attorneys using these tools are, in the rule's words, "required to carefully review the paper and independently ensure that it contains no fabricated or fictitious cases, statutes, or other material." That is the sentence to tape to your monitor. It does not ask whether you used AI. It asks whether what you filed is real.

The rule does not float free of existing law. It is anchored in 22 NYCRR Section 130-1.1 and Rule 3.3 of the New York Rules of Professional Conduct, the provisions that already bar frivolous arguments and false material statements to a tribunal. In practical terms, Part 161 is less a new standard than a pointed reminder that the old standards apply to AI output exactly as they apply to a junior associate's memo.

One boundary worth marking: Part 161 addresses papers submitted to the court, such as briefs and affidavits. It does not expressly address AI used to create evidentiary material, which remains subject to existing rules of evidence and procedure. Do not assume this rule authorizes AI-generated exhibits.

Why "disclosure rule" is a misnomer

The disclosure label sticks because New York almost went the other way. A pending bill, Senate Bill S2698, would require attorneys to disclose AI use and certify that a human reviewed and verified the content. That is a genuine disclosure-and-certification model, and it is not the law. It is a bill. Part 161, the rule that actually took effect, chose the lighter path: no mandatory disclosure, no separate AI form, no checkbox.

What Part 161 leans on instead is implicit certification. Under the rule's approach, your signature on the document is your certification that you did the verification work. There is no new artifact to file. There is only a higher-stakes meaning attached to the signature you were already putting on every paper.

So the honest one-line version is this: Part 161 is the AI rule that declines to make you announce your AI. It makes you answer for it.

The judge-by-judge catch that changes your prep

Here is where litigators get burned if they read only the headline. Part 161 is binding statewide. It is the mandatory floor, adopted by the Chief Administrator of the Courts, and no judge gets to opt out of it. What individual judges can do is go further: a judge may layer stricter AI requirements into their own part rules on top of Part 161.

That means the statewide floor is fixed, but the ceiling varies by courtroom. A judge in one part may demand an affirmative AI-use statement in every filing; a judge down the hall may add nothing beyond what Part 161 already requires. Both are consistent with the rule, because the extra requirements only ever stack on top of the mandatory baseline.

For your workflow, this collapses into one habit: check the assigned judge's individual part rules before you file, every time, the same way you already check page limits and formatting. The rule that governs your specific filing is Part 161 plus whatever that judge has layered on top.

What happens when the citations are fake

Part 161 does not invent a new penalty. It routes you back to the sanctions machinery that already exists. A single AI-generated falsehood in a filing can expose you to sanctions, fee-shifting, and potential disciplinary referral under the professional conduct rules the state has enforced for years.

That continuity is the point. Courts across the country have already sanctioned lawyers for filing briefs with invented cases produced by chatbots, and New York's other recent moves have hardened that posture. If you want the appellate-court version of how badly this can go, see our reporting on [New York appellate AI hallucination sanctions](/ai-regulation-news/new-york-appellate-ai-hallucination-sanctions-2026). Part 161 simply removes any argument that the lawyer did not know the rules covered AI. Now everyone knows.

How to comply without slowing down

Verification is the only affirmative duty the rule imposes, so build your process around it. Here is a checklist you can adopt today:

• Treat every AI draft as an unverified draft. Nothing an AI tool produces is citable until a human confirms it.
• Pull each authority to primary source. Every citation, every case reference, every statutory quote gets confirmed against the actual reporter, statute, or docket, not against the model's summary.
• Verify the holding, not just the existence. A case can be real and still be described wrong. Read enough to confirm the proposition you are citing it for.
• Check the judge's part rules for AI-specific requirements before filing, and log what you find so your team is not re-checking blind.
• Keep a short verification note in the file. You are not required to file it, but if a citation is later challenged, a contemporaneous record that a human checked the source is worth having.
• Supervise the tool like staff. The duty to review does not shift because a machine drafted the language. The signing attorney owns it.

If you want a decision rule for the gray cases, use this: would you file this exact sentence if a first-year associate had written it and you had not checked their sources? If the answer is no, you are not done, regardless of how confident the AI sounded.

The bigger shift for New York practice

Part 161 is best understood as New York picking a philosophy. The state could have made AI a reporting event, something you flag and certify on the record. It chose instead to make AI a competence event, something you are responsible for whether or not anyone asks. That is a more demanding standard in practice, because it never goes away and it never gets satisfied by a form.

For litigators, the winners under this rule are the ones who already treat AI output as a fast first draft from an unreliable clerk. The exposure sits with anyone still pasting model output into a brief and trusting the citations. The rule does not slow AI adoption. It raises the price of using it carelessly. Lawyers who want to build that verification discipline into a real workflow can start with our [Leveraged Attorney course](/leveraged-attorney), and if you are not sure where your AI skills stand, the [two-minute quiz](/quiz) is a fast gut check.

Browse the full AI Regulation News tracker

Informational analysis for working professionals, not legal advice. Confirm how any rule applies to your situation with qualified counsel.