AI and client confidentiality: what New York lawyers need to know

If you practice in New York and use ChatGPT, Claude or another AI tool, three documents set the ground rules, and one court rule sits next to them. This is the short version for a solo or small-firm lawyer. It is based on the documents themselves, listed at the bottom.

The rule underneath: Rule 1.6(c)

New York's Rules of Professional Conduct already cover this. Rule 1.6(c) says:

"A lawyer shall make reasonable efforts to prevent the inadvertent or unauthorized disclosure or use of, or unauthorized access to, information protected by Rules 1.6, 1.9(c), or 1.18(b)."

Typing client information into an AI tool is one more place that duty applies. The two bar documents below explain how.

NYC Bar Formal Opinion 2024-5

The New York City Bar Association published this opinion on August 7, 2024. It covers lawyers' use of generative AI. On confidentiality, the core line is:

"Without client consent, a lawyer must not input confidential client information into any Generative AI system that will share the inputted confidential information with third parties."

What else it says, in plain words:

• Open and closed tools are different. A tool that shares what you type with third parties is what the opinion calls "open." A tool that keeps inputs entirely inside the firm's own protected systems is "closed," and the opinion says it does not present the same risks.
• Closed still needs care. Even inside the firm, client information should not reach other people or other clients of the same firm who should not see it.
• Even with consent, leave out identifying details. The opinion says a lawyer should "avoid entering details that can be used to identify the client."
• Anonymizing can remove the need for consent. In its words: "Consent is not needed if no confidential client information is shared, for example through anonymization of client information."
• Read the terms of use, and keep reading them. Terms change. The opinion tells lawyers to check them from time to time.
• Consent has to be informed. The client must understand what sharing could lead to.
• Supervisors set the policy. Managing lawyers should set clear rules for AI use and train their people.
• Billing. You must not bill hourly for time the AI saved you.

It is a bar association opinion. It interprets the rules. It is not a court rule.

The NYSBA Task Force report

The New York State Bar Association's Task Force on Artificial Intelligence issued its report in April 2024. The NYSBA House of Delegates adopted it on April 6, 2024. It is a report with proposed guidelines, not a formal ethics opinion.

It flags the same risk: confidentiality concerns arise when you enter information into AI tools and those entries are added to the tool's training data. It also says consent is not the end of it:

"Even if your client gives informed consent for you to input confidential information into a Tool, you should obtain assurance that the Tool provider will protect your client's confidential information"

Part 161: a court rule about filings, not client data

22 NYCRR Part 161, "Use of Artificial Intelligence Technology," was adopted by administrative order on March 25, 2026. It took effect June 1, 2026. Here is what it does:

• Covers all courts of the Unified Court System, civil and criminal.
• No disclosure duty by default. Section 161.3 says lawyers and parties should not be required to tell the court they used AI on their papers.
• A judge may adopt a part rule. Courts are encouraged to use a model rule in Appendix A. Under it, anyone who used AI on a paper must carefully review it and make sure it has no made-up cases, statutes or other material. Signing the paper counts as your statement that you did that review.
• Silent on client data. Part 161 says nothing about confidentiality or what you put into an AI tool. That question stays with Rule 1.6 and the bar guidance above.

For context, the court system's own AI policy for judges and court staff, effective October 2025, takes a hard line on public tools. It says to assume that anything entered into a public AI platform such as ChatGPT "will immediately become public." That policy binds court employees, not lawyers. But it shows how the courts see these tools.

Recording calls with AI

The NYC Bar has two newer opinions on AI note-takers. Formal Opinion 2025-6 (December 22, 2025) says to get client consent before recording a client call. Formal Opinion 2026-2 (August 5, 2026) says to get every other party's consent before recording a call with non-clients.

Five steps for New York lawyers

1. List the AI tools you and your staff use. Include note-takers, browser extensions and AI built into other software.
2. Sort them into open and closed. Read each tool's terms. If it shares inputs with third parties, or you cannot tell, treat it as open.
3. For open tools, choose consent or anonymize. Get the client's informed consent, or take the client details out so no confidential information goes in. Even with consent, leave out details that identify the client.
4. Check the output. Verify every case and statute yourself. If the judge has adopted a Part 161 part rule, follow it.
5. Write a firm policy. Say which tools are allowed, what may go into each, and who checks the work. Train your staff on it.

How to do step 3 with KeptPDF

Anonymize for AI looks for names, dates of birth, Social Security numbers, driver's license numbers, account numbers, addresses and phone numbers. It swaps each one for a numbered placeholder like [PERSON_1]. It runs in your browser. The file is not uploaded to KeptPDF.

It does not find everything. It does not read medical or psychiatric details, criminal ID numbers or sealed material out of your text. It does not know which case facts are confidential. Facts like an employer, a date and a small town can point to a client with every name gone. You review those yourself.

1. Open Anonymize for AI and drop in a PDF or Word file, or paste text.
2. Review what it found. Add anything it missed, including case facts that point to the client.
3. Copy the anonymized text into ChatGPT, Claude or whatever you use.
4. Put the real names back. Paste the AI's answer into the restore box. Keep the tab open until you have done this.

One honest note. The cleaned text you paste does go to the AI company. That is the point. The details you took out stay behind. If a detail does not need to reach anyone, the Redact tool removes it from the PDF for good.

Sources

Every source below is the issuer's own document. Outside the short quotes above, we summarize in our own words.

• New York Rules of Professional Conduct, 22 NYCRR Part 1200 (nycourts.gov)
• NYC Bar Formal Opinion 2024-5: Ethical Obligations of Lawyers and Law Firms Relating to the Use of Generative Artificial Intelligence in the Practice of Law (August 7, 2024)
• NYSBA: Report and Recommendations of the Task Force on Artificial Intelligence (April 2024)
• NY Courts: Part 161, Use of Artificial Intelligence Technology
• NY Courts: Administrative Order AO/75/2026 adopting Part 161 (March 25, 2026)
• NY Unified Court System: Interim Policy on the Use of Artificial Intelligence (effective October 2025)
• NYC Bar Formal Opinion 2025-6 (December 22, 2025)
• NYC Bar Formal Opinion 2026-2 (August 5, 2026)
• ABA Formal Opinion 512 (July 29, 2024), the national baseline. See our ABA 512 guide.

This is not legal advice. It is a plain-English summary for general information. Talk to your own ethics counsel about your practice.