New York State Bar Adopts AI Ethics Guidelines

The New York State Bar Association’s House of Delegates adopted a report titled “Legal Ethics Guidelines for Lawyers’ Use of Artificial Intelligence” on Saturday, October 31. The guidelines come less than a week after New York’s court system published its own AI guide for attorneys, so New York lawyers now have two fresh documents telling them how to use these tools, one from the courts and one from the bar.

The report was written by the association’s Committee on Artificial Intelligence and Emerging Technologies, chaired by Vivian Wesson, general counsel and chief compliance officer of The Board of Pensions of the Presbyterian Church (U.S.A.). It is guidance from a voluntary bar association, not a change to the New York Rules of Professional Conduct, but it reads those rules onto AI use in detail.

What the Guidelines Say

According to the association’s announcement, the report tells lawyers to keep legal decisions in human hands, avoid the unauthorized practice of law, and check that AI results do not reflect biased training data, especially where a tool could affect a person’s rights, access, opportunity or treatment.

It also covers conflicts of interest when a tool uses data about a party in a case, notifying all parties when AI is used to record and summarize, disclosing AI use in court submissions, and reviewing the output of agentic systems that complete tasks without human input.

“The guidelines can help attorneys avoid possible perils of using AI including inadvertently practicing law in a jurisdiction where they are not licensed or failing to notify clients or the courts of AI use,” said NYSBA President Taa Grays.

In her address to the House of Delegates, Grays summed up the report’s central message: “we are fully responsible for ensuring that any AI output we incorporate into our work is accurate and secure.”

Billing and Vendor Rules Firms Will Feel First

The most practical sections deal with money and supervision. The association says lawyers may charge for time spent verifying an AI tool’s output, but not for time spent learning how to use the tool. Fees must stay reasonable and specific charges should be disclosed to clients. The cost of acquiring an AI tool is treated as overhead, so it may not be billed to a client unless it is a service specially incurred for that matter.

On supervision, the report notes that the New York rules do not specifically address AI-generated legal work. It extends the existing duty to supervise non-lawyers, so a lawyer remains responsible for the quality of work produced by staff and vendors using AI, and by the AI tool itself.

For firm leaders and legal ops teams, that points at engagement letters, billing guidelines, and vendor contracts. A firm that bills “AI review” time or passes through software costs now has a published bar view to measure those practices against.

How It Fits With the Courts’ Guide

On October 26, New York’s Unified Court System released a seven-page guide, The Risks of Using Artificial Intelligence in Preparing Court Papers: What Every Attorney Should Know, as reported by the ABA Journal. Law.com reported that the guide builds on a statewide rule adopted earlier this year that puts responsibility for verifying AI-assisted filings on lawyers.

The two documents push in the same direction on verification. They differ in emphasis on disclosure: the court guide centers on checking work, while the bar’s report lists disclosing AI use in court submissions among its topics, describing it as a policy many courts already have in place. Lawyers still need to follow the specific rules of the court where they file.

The ABA Journal also noted that the ABA’s ethics committee addressed lawyers’ AI duties in a 2024 formal opinion, and that Rhode Island and Ohio courts have issued their own guidance. New York’s bar report adds a detailed state-level layer on top.

Common Questions

Are the NYSBA guidelines binding?

They are a bar association report, not a court rule or an amendment to the Rules of Professional Conduct. They explain how the association reads existing rules for AI use.

Can a firm bill for AI verification time?

According to the association, time spent verifying AI output may be billed, but time spent learning a tool may not, and tool costs are generally overhead.

Do the guidelines cover AI agents?

Yes. The report addresses reviewing the output of agentic AI systems.

What to Watch

Watch for the full report text to circulate, whether the Appellate Divisions or other state bars borrow its billing and supervision positions, and how vendors selling to New York firms answer questions about bias testing and agent oversight.