Research it yourself

Why this matters

This guide is for employees bringing a leave, disability, or retaliation claim, most often without a lawyer. If you are representing yourself, you will have to find and read the law on your own. An AI tool can make that faster and cheaper. It can also hand you a case that does not exist, and a judge or arbitrator will hold you responsible for it. This guide shows a method that keeps the speed and removes most of the risk.

This is general information, not legal advice.

What AI does well

Used with care, an AI tool is a fast assistant. It is good at four jobs.

• Finding leads. It can suggest search terms, statutes, and the kinds of cases that might help you. Treat every lead as a guess until you read the source.
• Reading fast. You can give it a long opinion or a stack of emails and ask for a summary, a timeline, or the passages that mention a topic.
• Organizing. It can sort documents, build a chronology, and keep a list of what you have checked.
• Drafting. It can turn your notes into a clear first draft that you then check line by line.

What AI does badly

The same tools fail in ways that are easy to miss because the output sounds sure of itself.

• Inventing cases. A tool can produce a case name, a court, a date, and a quote that look real and are not.
• Misstating holdings. A real case can be described as holding the opposite of what it holds, or as holding something it never reached.
• Confident gaps. A tool often answers a question it cannot answer. It rarely says "I could not find this."
• Stale law. A tool may not know that a case was overruled or that a rule changed.

The one rule

Never cite a case you have not read in full.

That means you opened the actual opinion from a trustworthy source, read it, and found the passage you rely on. A case name in an AI answer, a blog post, or another court's opinion is a lead. It is not a verified cite. Chapter 6 gives the steps.

Real consequences

Courts have already sanctioned or disciplined people for filing fake citations. The three decisions below were read in full for this guide.

Mata v. Avianca

In Mata v. Avianca, Inc., 678 F. Supp. 3d 443, 2023 WL 4114965 (S.D.N.Y. 2023), two lawyers and their firm filed cases that a chatbot had made up. They kept defending the filing after the court and the other side raised questions. The court found subjective bad faith and imposed a $5,000 penalty on the lawyers and firm jointly, and it required letters to the client and to each judge falsely named as an author. The court also said "there is nothing inherently improper about using a reliable artificial intelligence tool for assistance." But existing rules give attorneys a gatekeeping role to make sure their filings are accurate.

Park v. Kim

In Park v. Kim, 91 F.4th 610 (2d Cir. 2024), the Second Circuit referred a lawyer to its Grievance Panel. Her reply brief cited a nonexistent state case that ChatGPT had suggested, and she did not read or otherwise confirm it. The court said this fell well below the basic obligations of counsel. It then quoted Mata, the district court decision above, for this point. "A fake opinion is not 'existing law' and citation to a fake opinion does not provide a non-frivolous ground for extending, modifying, or reversing existing law, or for establishing new law. An attempt to persuade a court or oppose an adversary by relying on fake opinions is an abuse of the adversary system."

Matter of Julien v. Arthur

This one involves someone without a lawyer. In Matter of Julien v. Arthur, 2026 NY Slip Op 03308 (2d Dep't May 27, 2026), a self-represented father filed an appeal brief that cited a nonexistent case, apparently produced by AI. The court held that "the unverified usage of GenAI to draft an appellate brief containing false information constitutes frivolous conduct warranting the imposition of a sanction, even when the offending party is a pro se litigant." It imposed a $250 sanction. In setting that amount, the court noted that he cited one fake case rather than many, did not repeat the error, and accepted responsibility. It warned that future misuse could bring a greater sanction. This is a slip opinion, so check for the final published cite.

Not having a lawyer is not a defense. It may soften the penalty, but it does not remove your duty to check.

Court rules on AI

Courts are writing rules about AI. New York is one example. Its statewide policy, 22 NYCRR Part 161, took effect June 1, 2026. It says attorneys and parties should not be barred from using AI tools to prepare court papers, and should not be required to disclose that they did, because the duties that apply to every filing already apply. A court may adopt its own part rule on AI, and if it does, it is encouraged to use a model rule attached to Part 161. That model rule says a person who uses an AI tool must carefully review the paper and independently ensure it contains no fabricated or fictitious cases, statutes, or other material. Signing the paper certifies that the review was done, and a failure may bring a sanction. Whether your court has adopted a rule varies, so check your assigned judge's part rules.

Part 161 does not cover federal courts. Federal judges' standing orders vary, and arbitrators may set their own rules. Chapter 10 covers how to check yours.

How this guide is organized

What this method does not do

It does not replace a lawyer's judgment. It does not give you a paid citator. It does not make a bad cite impossible. It makes one detectable, because nothing counts as verified until you have read and saved the opinion. You stay responsible for everything you file.

Next: Chapter 1, what you need to get started.

Sources

• New York 22 NYCRR Part 161, Use of Artificial Intelligence Technology
• Administrative Order AO/75/2026 (PDF)
• CourtListener