New York: AI Ethics Guidance for Law Firms

Bottom line for a 5-50 attorney New York firm: Three NYC Bar formal opinions, the NYSBA Task Force Report, and 22 NYCRR Part 161 (in force since 2026-06-01) together describe the expected practice on vendor vetting, citation verification, supervision, and consent to AI recording. Part 161 sets a system-wide policy that attorneys should not be required to disclose AI use, and it lets individual courts adopt an optional Model Rule (Appendix A) under which a lawyer who used AI must review the paper for fabricated content and signing certifies that review; firms may wish to record, for each court and judge they appear before, whether that Model Rule or a stricter part rule applies. The NYC Bar opinions conclude that clients must be notified, and consent obtained, before AI records their calls, and that the default should be not to record calls with non-clients. The biggest exposure is a stricken filing or sanctions referral arising from an unverified AI citation, which creates a direct line to malpractice exposure regardless of any specific disclosure rule.

Start with: a written AI use policy, a court-filing verification protocol, and a vendor due-diligence record for each AI tool that touches client data.

NYC Bar Formal Opinion 2024-5

Citation: New York City Bar Association Committee on Professional Ethics, Formal Opinion 2024-5 (issued August 7, 2024). Primary source PDF. Landing page.

Status: Formal advisory opinion of a voluntary bar association. The Committee’s formal and informal opinions “provide general guidance to New York lawyers” (Committee page); they are not binding.

What it says lawyers must or should do

• Competence. “Before selecting and using a Generative AI tool, a lawyer should understand to a reasonable degree how the technology works, its limitations, and the applicable [T]erms of [U]se.” The opinion adds that using such tools “without the application of trained judgment by a lawyer is inconsistent with the competent and diligent practice of law.”
• Confidentiality. “[A] lawyer must not input any confidential information of the client into any Generative AI system that lacks adequate confidentiality and security protections,” unless the client has given informed consent. A lawyer “should review the system’s Terms of Use” and monitor them over time.
• Conflicts. “Where a Generative AI system uses client information, a law firm must ensure that the system implements any ethical screens required under the Rules,” and even with closed systems a lawyer “must take care that confidential information is not improperly shared with other persons at or clients of the same law firm.”
• Supervision. “Managerial and supervisory lawyers should establish clear policies regarding the permissible uses of [G]enerative AI.”
• Verification and candor. “A lawyer must review all [G]enerative AI outputs” for accuracy before using them for client purposes or submitting them to a court, and “must correct any errors or misleading statements made to adversaries, the public, or the court.” The opinion addresses these duties under Rules 3.1 and 3.3.
• Fees. “A lawyer must not charge hourly fees for the time that would otherwise have been spent absent the use of Generative AI”; costs associated with generative AI “should be disclosed in advance to clients as required by Rule 1.5(b).”

Client disclosure

The opinion does not require disclosure in every case. A lawyer “should consider disclosing to the client the intent to use Generative AI that is not generally understood to be routinely used by lawyers as part of the representation,” “should obtain client consent for Generative AI use if client confidences will be disclosed,” and “should also check for any rules, orders, or other requirements in the relevant jurisdiction that may necessitate the disclosure.”

Notable gaps

The opinion does not identify which tools meet its confidentiality standard and does not address agentic AI systems. It describes itself as general guidance that will be updated and supplemented.

NYC Bar Formal Opinion 2025-6

Citation: New York City Bar Association Committee on Professional Ethics, Formal Opinion 2025-6 (issued December 22, 2025), on AI tools that record, transcribe, or summarize conversations with clients. Primary source.

Status: Formal advisory opinion; not binding. It cites the Preamble and Rules 1.1, 1.6, 5.1, 5.3, and 8.4.

What it says

• Notice and consent. Grounding the analysis in the duty of loyalty and Rule 8.4, the opinion concludes that “clients must be notified, and their consent obtained, whenever their calls are being recorded by an AI-empowered system.” It notes that New York is generally a one-party consent state under its recording laws, but treats the ethics question separately. Where a platform such as Zoom or Teams does not give clear notice that a call is being recorded, the client should be notified, and consent obtained, each time.
• Competence. “[A]ttorneys have a duty to acquire an understanding of the technical features of the AI tools they and their clients use.”
• Confidentiality. Attorneys “should consider what privacy and security safeguards are in place in an AI tool to protect the data, including where data will be stored and for how long, how data might be retrievable through discovery, whether the tool uses such data for training, and whether there is a right to data deletion.”
• Accuracy. An attorney “should check the work product for accuracy if there is a possibility the transcription or summary will be preserved and potentially relied upon.”
• Client-side recording. If an attorney knows a client is recording a call with an AI tool, the lawyer should advise the client of the disadvantages of doing so. The opinion suggests that “the ideal solution is for clients to agree not to use their own AI tools to record conversations with attorneys without advance notice,” and that retainer agreements may address the status of client-made recordings.
• Retention. A lawyer or firm that adopts a policy not to retain recordings and transcripts should disclose that policy.
• Biometric data. Some states’ biometric privacy laws treat voiceprints as protected information.

Notable gaps

The opinion does not identify which recording or transcription vendors meet its safeguards.

NYC Bar Formal Opinion 2026-2

Citation: New York City Bar Association Committee on Professional Ethics, Formal Opinion 2026-2 (issued August 5, 2026), “Ethical Issues Affecting Use of AI to Record, Transcribe, and Summarize Conversations with Persons Who Are Not Clients.” Primary source.

Status: Formal advisory opinion; not binding. It cites Rules 1.1, 1.18, 3.4, 4.2, 4.3, and 8.4.

What it says

• The opinion’s digest: “an attorney should obtain consent of all other parties to a call before recording it, and should consider whether recording, transcribing and summarizing is tactically well-advised in the particular circumstances.”
• Restating Opinion 2025-6 for third parties: “attorneys must not surreptitiously record conversations, including videoconferences, with any third parties. Attorneys must disclose the intention to record the conversation and obtain permission from all participants.”
• “The duty of competence includes knowing how to turn off the recording function.”
• Default positions by context. For prospective clients (Rule 1.18), the default practice should be not to record. For witnesses (Rules 3.4, 4.2, 4.3), “[b]ecause the potential risks ordinarily outweigh the benefits, the default practice should be not to record the conversations.” For co-counsel, the opinion suggests addressing recording in any joint defense or common interest agreement. Conversations with opposing counsel raise settlement and mediation-privilege concerns.

NYSBA Task Force on Artificial Intelligence: Report and Recommendations

Citation: NYSBA Task Force on Artificial Intelligence, Report and Recommendations, approved by the NYSBA House of Delegates April 6, 2024. Primary source PDF. NYSBA’s standing work on AI now sits with its Committee on Artificial Intelligence and Emerging Technologies.

Status: Task force report with adopted recommendations. Non-binding guidance, not a formal ethics opinion or rule amendment.

What it says (per the cited rules)

• Lawyers have “a duty to understand the benefits, risks and ethical implications … associated with the Tools” (Rule 1.1).
• Attorneys “must verify the accuracy of the information and legal authority produced by such tools.”
• On confidentiality (Rule 1.6), lawyers “must take precautions to protect sensitive client data,” should obtain assurance that the tool provider will protect the client’s confidential information, and should periodically monitor the provider.
• A supervising lawyer has “a duty to ensure that the lawyers for whom you have oversight observe the ethical rules when utilizing the Tools” (Rules 5.1 and 5.3).

What it recommends

• Consider including in the engagement letter a statement that AI tools may be used in the representation, and seek the client’s acknowledgement (sample language appears in the report’s Appendix C).
• On fees, the report notes that if the tools “would make your work on behalf of a client substantially more efficient, then your use of (or failure to use) such Tools may be considered as a factor in determining whether the fees you charged for a given task or matter were reasonable.”
• Expand Comment [8] to Rule 1.1 to address AI. As of NYSBA’s 2025 edition of the Rules (comments as amended through January 1, 2025), Comment [8] had not been amended to mention AI (NYSBA 2025 edition).

22 NYCRR Part 161

Citation: Rules of the Chief Administrator of the Courts, Part 161 (22 NYCRR Part 161), added by Administrative Order AO/75/26 of the Chief Administrative Judge, dated 2026-03-25 and effective 2026-06-01. Primary source. AO/75/26 transmittal letter. Tracker entry: 22 NYCRR Part 161.

Status: In force since 2026-06-01 in “all courts of the Unified Court System, in both civil and criminal cases” (Section 161.1). It is a statewide policy, not a new filing obligation: the review-and-certify duty described below applies only in a court that adopts the Model Rule.

What it provides

• Section 161.3 (policy): attorneys’ and parties’ use of AI tools in preparing court papers “should not be prohibited” when it accords with the duties that already apply to submissions, and attorneys and parties “should not be required, upon submitting papers, to disclose to the court that they have used AI in the preparation of such papers.”
• Section 161.4 (Model Rule): a court “may, in its discretion, implement a part rule” on AI use in preparing papers and, if it does, “is encouraged to adopt the model rule set forth in Appendix A.”
• Appendix A (Model Rule): in a court that adopts it, an attorney or party who uses an AI tool in preparing a paper filed, submitted, or served in a case before that court “is required to carefully review the paper and independently ensure that it contains no fabricated or fictitious cases, statutes, or other material.” Signing the paper certifies that the review was conducted, and a court that finds the requirement unmet may impose “sanction or other remedial action.”
• Section 161.2 (definitions): “artificial intelligence” is defined in machine-based-system terms, and “paper” covers briefs, memoranda, affidavits, affirmations, pleadings, and similar documents, but excludes materials “constituting or proffered as evidence in the case.”
• The Model Rule itself notes that existing certification duties under 22 NYCRR 130-1.1 and 130-1.1a and the Rules of Professional Conduct apply to every paper, whether or not AI was used.

Notable gaps

Part 161 does not require disclosure, so a court applying only the statewide policy has no AI-specific record of which filings used AI. Whether a given court or part has adopted the Model Rule, or a stricter part rule of its own, has to be checked court by court and judge by judge; firms may wish to record that check at case intake.

What federal courts in New York require for AI use in filings

None of New York’s four federal districts has a district-wide AI local rule: the joint S.D.N.Y. and E.D.N.Y. Local Rules (effective January 2, 2026), the N.D.N.Y. Local Rules (effective January 1, 2026), and the W.D.N.Y. Local Rules of Civil Procedure (effective January 15, 2026) contain no AI provision. The operative federal default is Fed. R. Civ. P. 11. Two exceptions matter:

• Bankruptcy Court, S.D.N.Y.: Local Bankruptcy Rule 9011-1(d) (effective September 30, 2024) provides that litigants “remain responsible for the accuracy and quality of legal documents produced with the assistance of technology” and “must review and verify any computer-generated content” under Fed. R. Bankr. P. 9011. It requires verification, not disclosure (Local Bankruptcy Rules).
• Individual judges’ practices. Several S.D.N.Y. and E.D.N.Y. judges address AI in their individual rules; the list below is not exhaustive. Judge Broderick (S.D.N.Y.) requires any party that uses a generative AI tool in preparing a filing to disclose that use (entry). Judge Cronan (S.D.N.Y.) requires a signed certification describing the steps taken to verify AI-generated authorities and factual assertions (entry). Judge Ho (S.D.N.Y.) requires a separate declaration disclosing AI use in civil, civil pro se, and criminal filings (entry). Magistrate Judge Lindsay (E.D.N.Y.) requires disclosure of AI use and certification that AI-drafted portions were checked (entry), and Magistrate Judge Dunst (E.D.N.Y.) cautions parties that AI-assisted submissions must still comply with their professional obligations (entry). Firms may wish to check the assigned judge’s individual practices at the opening of each federal matter.

What New York state courts require for AI use in filings

Part 161 sets the policy for every court of the Unified Court System. Individual justices’ part rules still vary, and several predate Part 161:

• Justice Maslow (Kings County Supreme Court) requires all submissions on a motion to include a certification that either no generative AI program was used or identifies the program and the AI-drafted portions, and requires an attorney who discovers hallucinations in an opposing submission to inform the court (entry).
• Justice Dawson (New York County Supreme Court, Integrated Domestic Violence Part) requires an affirmation at the end of any document involving substantive legal research and writing “stating that AI, such as ChatGPT, was used in preparation of the document” (entry).
• Justice Norris (Niagara County Supreme Court) replaced his earlier four-part disclosure rule with the text of the Part 161 Model Rule (entry), and Justice Hanlon (Chautauqua County Supreme Court) now refers filers to Part 161 (entry).

Requirements vary by judge, so firms may wish to confirm the assigned judge’s current part rules at case intake.

On 2025-10-10, the New York State Unified Court System also issued an Interim Policy on the Use of Artificial Intelligence (press release). “The policy applies to all UCS judges, justices, and nonjudicial employees” and mandates AI training for them; it does not govern practicing attorneys, but it means New York judges work under an AI governance framework of their own. The Advisory Committee on Artificial Intelligence and the Courts published an Annual Report in December 2025.

What AI-related rules are pending in New York?

Senate Bill S2698 (Hoylman-Sigal; introduced January 22, 2025) would add CPLR 2107, under which “[a]ny paper or file served that was drafted with the assistance of generative artificial intelligence must attach to the filing a separate affidavit disclosing such use and certifying that a human being has reviewed the source material and verified that the artificially generated content is accurate,” with a parallel disclosure for appellants’ briefs under CPLR 5528 (bill text). It applies to any filer, not only attorneys. The Senate Judiciary Committee reported it in May 2025 and it advanced to third reading, but it was committed to the Senate Rules Committee on June 13, 2025, where it remains. An identical bill, S9794 (introduced April 6, 2026), followed the same path to Senate Rules in June 2026, and the Assembly companion, A8546, is in the Assembly Judiciary Committee. If enacted, S2698 would go further than Part 161, which states that disclosure should not be required. A prior version died in committee in 2023-2024.

Separately, S7263 (Gonzalez; introduced April 7, 2025; amended as S7263A on May 26, 2026) would bar a chatbot operator from knowingly permitting its chatbot to impersonate a licensed professional in a way that, if done by a person, would violate the Judiciary Law’s prohibition on unauthorized practice of law, with Attorney General enforcement. It is in the Senate Rules Committee. It is not attorney conduct legislation, but it signals legislative attention to AI and legal services.

The NYSBA Task Force’s recommendation to expand Comment [8] to Rule 1.1 to address AI had not been reflected in the Rules’ comments as of NYSBA’s 2025 edition.

CLE requirements

As of September 2026, New York does not have a standalone AI-specific CLE requirement. Experienced attorneys “must complete at least 1 CLE credit hour in Cybersecurity, Privacy and Data Protection as part of their biennial CLE requirement” (CLE Board news). Newly admitted attorneys must complete a minimum of 32 credit hours across their first two years, including at least one credit hour of cybersecurity, privacy and data protection (22 NYCRR 1500.12). Some providers offer AI programs for credit in that category (see the NYSBA cybersecurity CLE page).

How do malpractice carriers in New York treat AI use?

No New York bar opinion explicitly addresses malpractice carrier obligations. Of the LPL application forms reviewed for this site, only AmTrust’s asks about AI, and broker commentary reports LPL underwriters focusing on firms’ AI controls. No filed AI exclusion on an admitted-market LPL form has been located; the “absolute” AI exclusion form W.R. Berkley issued (Form PC 51380) sits on management liability, not LPL. NYSBA operates a malpractice insurance program through USI Affinity at nysbainsurance.com, which has not published AI-specific underwriting guidance. NY DFS Circular Letter No. 7 (July 11, 2024) issued guidance on AI use in insurance underwriting and pricing; it applies to carriers, not attorneys, but signals regulatory attention to AI in the insurance market.

What does my New York malpractice carrier ask about AI at renewal?

Dominant New York carriers have not published AI-specific application items the research note documents. Five substantive sources apply: NYC Bar Opinions 2024-5, 2025-6, and 2026-2, the NYSBA Task Force recommendations, and Part 161. Documentation should map to vendor vetting (per Op 2024-5), citation verification (Rule 3.3 plus Part 161), supervision under Rules 5.1 and 5.3, and recording consent under Opinions 2025-6 and 2026-2 where applicable. Part 161 has been in force since 2026-06-01, so a verification protocol for New York court filings is already part of the expected record.

What documentation should a New York firm keep on file?

Month one (foundational)

1. (Owner: managing partner + firm administrator) Written AI use policy covering approved tools, what categories of client information may or may not be entered into each tool, supervision responsibilities, and violation procedures. NYC Bar Opinion 2024-5 says managerial and supervisory lawyers “should establish clear policies” on AI use (Rules 5.1 and 5.3).
2. (Owner: litigation lead) Court filing verification protocol requiring any AI-assisted submission be reviewed specifically for hallucinated citations, fabricated case names, inaccurate quotations, and fabricated statutes. It supports Rule 3.3, Rule 130-1.1, and, in courts that have adopted it, the Part 161 Model Rule.
3. (Owner: matter lead attorney) Judge-specific AI part rule log checked at case assignment for each judge’s individual AI disclosure and use requirements, since Part 161 leaves those rules to individual courts.

Months two and three (operational documentation)

1. (Owner: firm administrator + outside IT) Vendor due diligence file for each AI tool that receives client data: data retention policy, training data practices, security certifications, and data processing agreement. Supports the confidentiality guidance in NYC Bar Opinion 2024-5.
2. (Owner: managing partner + billing partner) Engagement letter AI disclosure clause. If the firm uses AI transcription, include recording consent or prohibition language. Sources include the NYSBA Task Force sample clause and NYC Bar Opinion 2025-6 on recording consent.
3. (Owner: firm administrator) CLE compliance record documenting that each attorney has completed the 1-credit Cybersecurity, Privacy and Data Protection CLE each biennial cycle.

Months four to six (per-matter discipline)

These are recurring practices, not one-time projects: each new matter exercises them.

1. (Owner: matter lead attorney) Recording and transcription consent documentation capturing client notice and consent before any AI-recorded or AI-transcribed client communication (NYC Bar Opinion 2025-6), and the consent of all participants before recording a call with a non-client (Opinion 2026-2), plus the vendor data-handling agreement.
2. (Owner: managing partner) AI incident or error log recording identified AI errors and how they were handled. Supports supervision obligations under Rules 5.1 and 5.3.
3. (Owner: managing partner) Periodic review schedule for updated NYC Bar and NYSBA guidance, S2698 status, new judge part rules, and updated vendor terms.

We are building practitioner resources for firms working through this list. The newsletter covers new resources as they ship.