NYC-BAR 2024-08-07

What are the New York ethics obligations of a lawyer or law firm using generative AI tools in practice (chatbots, legal-research AI, document review and analytics)?

Short answer: The opinion provides general guidance rather than hard rules. Per the opinion, a New York lawyer using generative AI must consider confidentiality (Rule 1.6), conflicts (Rules 1.7-1.12), competence and diligence (Rules 1.1, 1.3), advertising/solicitation (Rules 7.1, 7.3), the duty to comply with law (Rule 1.2(d)), supervision of lawyers and nonlawyers (Rules 5.1, 5.3), subordinate-lawyer duties (Rule 5.2), client consultation (Rule 1.4), candor to tribunals (Rule 3.3), non-meritorious claims (Rule 3.1), the limits on fees and costs (Rule 1.5), and the prohibition on discrimination (Rule 8.4).

Disclaimer: Advisory only. Not binding precedent.

About this page: The plain-English summary, reader guidance, and Q&A below were written by Ezel based on the official ethics opinion. The original opinion (linked at the bottom of this page) is the authoritative source for any reliance.

NYC Bar Formal Opinion 2024-5: Generative AI in Law Practice

Short answer: The opinion provides general guidance, not hard rules, on how New York lawyers using generative AI in practice should apply the New York Rules of Professional Conduct. Per the committee, lawyers using generative AI tools must consider the duty of confidentiality (Rule 1.6); the obligation to avoid conflicts of interest (Rules 1.7-1.12); the duties of competence and diligence (Rules 1.1, 1.3); the rules governing advertising and solicitation (Rules 7.1, 7.3); the duty to comply with law (Rule 1.2(d)); the duty to supervise lawyers and nonlawyers (Rules 5.1, 5.3); subordinate-lawyer duties (Rule 5.2); the duty to consult with clients (Rule 1.4); candor to tribunals (Rule 3.3); the prohibition on non-meritorious claims (Rule 3.1); the limitations on fees and costs (Rule 1.5); and the prohibition on discrimination (Rule 8.4). The opinion's view is that lawyers need "guardrails" rather than new restrictive rules.

Disclaimer: This is an advisory ethics opinion. Advisory opinions are not binding; they interpret the New York Rules of Professional Conduct and are persuasive authority. This summary is for research purposes only and is not legal advice. Verify current rules before acting on any specific guidance.

About this page: The plain-English summary and Q&A below were written by Ezel based on the official opinion. We do not reproduce the opinion text on this page; follow the linked source for the official text, which controls.

View original opinion

Plain-English summary

The opinion responds to the rapid expansion of generative AI tools used in legal practice: general-purpose chatbots, legal-research platforms with AI drafting and analysis features, and document-management vendors offering AI-assisted review and analytics. The committee acknowledges the guidance is general and that the summary of currently available tools will be quickly outdated.

The opinion follows the format of the California State Bar Standing Committee on Professional Responsibility and Conduct's November 2023 Practical Guidance for the Use of Generative Artificial Intelligence in the Practice of Law, and is in many places consistent with the California Guidance and the ABA, Florida, D.C., and New Jersey treatments cited in the opinion's footnotes. The committee's view is that lawyers in this developing area need guardrails rather than hard-and-fast restrictions or new rules that could stymie developments, and the opinion expressly disagrees with the New York State Bar Association Task Force's recommendation to adopt new rules specifically addressing generative AI.

The opinion's structural conclusion is the list of duties a New York lawyer must consider when using generative AI: confidentiality (Rule 1.6); conflicts (Rules 1.7, 1.8, 1.9, 1.10, 1.11, 1.12); competence and diligence (Rules 1.1, 1.3); advertising and solicitation (Rules 7.1, 7.3); compliance with law (Rule 1.2(d)); supervision of lawyers and nonlawyers (Rules 5.1, 5.3); subordinate-lawyer duties (Rule 5.2); client consultation (Rule 1.4); candor to tribunals (Rule 3.3); non-meritorious claims and contentions (Rule 3.1); limitations on fees and costs (Rule 1.5); and the prohibition on discrimination (Rule 8.4). The opinion treats this catalog as the framework rather than promulgating distinct AI-specific obligations.

On confidentiality, the opinion's footnotes flag that "open" generative AI systems that share inputted information with third parties present Rule 1.6 issues, and that client consent to use such systems must be knowing, citing N.Y. State Op. 1020 (lawyer may use cloud storage tools that do not provide reasonable protection only with informed client consent). On candor to tribunals and supervision, the opinion's footnotes reference Mata v. Avianca (S.D.N.Y. 2023) (sanctions for filings citing ChatGPT-hallucinated cases) and United States v. Cohen (S.D.N.Y. 2024) (lawyer admonished for citations the client obtained from Google Bard without verification), and cite a Stanford study finding hallucination rates of at least 75% for chatbot answers about core court rulings.

In practice

Under this opinion, a New York lawyer using generative AI in practice should evaluate the use against the full Rules framework rather than against any single AI-specific rule. Per the opinion, the framework rules include Rule 1.6 (confidentiality, especially for "open" systems), Rule 1.7-1.12 (conflicts), Rule 1.1 (competence in the technology and its limits), Rule 1.3 (diligence), Rule 1.4 (client consultation), Rule 1.5 (fees and costs), Rule 3.1 and 3.3 (non-meritorious claims and candor to the tribunal), Rule 5.1, 5.2, 5.3 (supervision and subordinate-lawyer duties), Rule 7.1 and 7.3 (advertising and solicitation), and Rule 8.4 (including the discrimination prohibition). The opinion specifically references hallucination risks under Rules 3.1 and 3.3, citing Mata v. Avianca and United States v. Cohen.

The opinion notes that not all generative AI uses require client disclosure: Microsoft Word's auto-complete and grammar functions, Westlaw, Lexis, and search-engine AI are given as examples of uses that the committee does not suggest require disclosure.

Common questions

Q: Does the opinion require lawyers to disclose generative AI use to clients?

A: The opinion does not impose a uniform disclosure rule. Per the opinion's footnote 21, the committee specifically declines to suggest that an attorney must disclose AI uses that are embedded in routine tools (Word, Westlaw, Lexis, search engines). For uses involving sharing of client information with "open" generative AI systems, the opinion (footnote 5) ties consent to Rule 1.6: client consent must be knowing, with the client understanding the potential consequences, citing N.Y. State Op. 1020 on cloud-storage analogues.

Q: What does the opinion say about lawyers being sanctioned for AI-hallucinated citations?

A: The opinion treats hallucination as a candor-to-tribunal and competence problem under Rules 3.1, 3.3, 1.1, and 1.3. The opinion's footnotes cite Mata v. Avianca, Inc., No. 22-CV-1461, 2023 WL 4114964 (S.D.N.Y. June 22, 2023) (sanctioning attorneys for submitting hallucinated citations) and United States v. Cohen, No. 18-CR-602, 2024 WL 1193604 (S.D.N.Y. Mar. 20, 2024) (counsel criticized for not verifying client-supplied Bard hallucinations).

Q: How does the opinion handle confidentiality for generative AI tools that share inputs with third parties?

A: The opinion (footnote 3) identifies systems that share inputted information with third parties as "open" systems, and (footnote 5) treats them as raising Rule 1.6 issues that require informed client consent before client information is shared. The opinion analogizes to cloud-storage practice under N.Y. State Op. 1020.

Q: Did the committee recommend new New York rules specifically addressing generative AI?

A: No. Per the opinion, the committee declines to recommend new AI-specific rules, taking the position that lawyers need guardrails rather than restrictions while the technology evolves. The committee specifically diverges from the New York State Bar Association Task Force, which suggested adoption of certain new rules.

Q: Does the opinion address intellectual-property risk in using generative AI?

A: The opinion (footnote 12, following the California Guidance) flags that some generative AI tools have been the subject of intellectual-property infringement claims, and instructs that a lawyer planning to use a tool should keep abreast of whether there are any such risks associated with the tool the lawyer plans to use.

Background and rules framework

The opinion is general guidance from the NYC Bar Professional Ethics Committee on the use of generative AI by New York lawyers, law firms, legal departments, government law offices, and legal assistance organizations. It interprets the existing New York Rules rather than creating new ones, and follows the California State Bar's November 2023 Practical Guidance. The opinion is consistent with ABA Formal Op. 512 (2024), Florida Bar Bd. Rev. Comm. on Pro. Ethics Op. 24-1 (2024), and D.C. Bar Ethics Op. 388 (April 2024), each cited in footnote 1. The opinion references N.Y. State Bar Ass'n Report & Recommendations of the Task Force on Artificial Intelligence (2024) for the contrast position that some new rules should be adopted.

Citations and references

Rules of Professional Conduct:

N.Y. RPC 1.1 (competence)
N.Y. RPC 1.2(d) (no assistance to crime or fraud)
N.Y. RPC 1.3 (diligence)
N.Y. RPC 1.4 (client communication)
N.Y. RPC 1.5 (fees and costs)
N.Y. RPC 1.6 (confidentiality)
N.Y. RPC 1.7, 1.8, 1.9, 1.10, 1.11, 1.12 (conflicts; current, former, imputed, government, former judicial)
N.Y. RPC 3.1 (non-meritorious contentions)
N.Y. RPC 3.3 (candor to tribunal)
N.Y. RPC 5.1, 5.2, 5.3 (supervision of lawyers; subordinate lawyers; nonlawyers)
N.Y. RPC 7.1, 7.3 (advertising and solicitation)
N.Y. RPC 8.4 (misconduct, including discrimination)

Cases:

Mata v. Avianca, Inc., No. 22-CV-1461, 2023 WL 4114964 (S.D.N.Y. June 22, 2023), sanctions for hallucinated citations.
United States v. Cohen, No. 18-CR-602, 2024 WL 1193604 (S.D.N.Y. Mar. 20, 2024), criticism for unverified client-supplied AI citations.

Other opinions cited:

California State Bar, Practical Guidance for the Use of Generative Artificial Intelligence in the Practice of Law (Nov. 16, 2023).
ABA Formal Op. 512 (2024).
Florida Bar Bd. Rev. Comm. on Pro. Ethics Op. 24-1 (2024).
D.C. Bar Ethics Op. 388 (April 2024).
N.J. State Bar Ass'n Task Force on AI and the Law (2024).
N.Y. State Bar Ass'n Report of the Task Force on Artificial Intelligence (2024).
ABA Formal Op. 491 (2020): duty to inquire when transaction may involve crime or fraud.
ABA Formal Op. 93-379 (1993): fees and costs.
Colo. Bar Ass'n Ethics Comm., Formal Op. 142 (2021).
NYC Bar Op. 2018-4: duty to inquire when asked to assist in a suspicious transaction.
N.Y. State Op. 1020: cloud storage and informed client consent.

Source

Landing page: https://www.nycbar.org/reports/formal-opinion-2024-5-generative-ai-in-the-practice-of-law/