Can a Pennsylvania lawyer use generative AI like ChatGPT in their practice?

Plain-English summary

This is a joint opinion of the Pennsylvania Bar Association Committee on Legal Ethics and Professional Responsibility and the Philadelphia Bar Association Professional Guidance Committee. It treats AI as raising the same ethical issues as other technologies, such as cloud computing, email, and smartphones, so existing Rules govern its use. The Committees conclude that "with appropriate safeguards, lawyers can utilize artificial intelligence in accordance with the Pennsylvania Rules of Professional Conduct."

On competence, the opinion states that "lawyers must be competent in the use of AI technologies": a lawyer who chooses to use AI must understand how it works, its benefits and risks, and must check and verify all citations and cited material. It stresses that generative AI can create content rather than merely analyze it, producing "hallucinations" and biases, and recounts that lawyers have been sanctioned for filing briefs citing nonexistent AI-generated cases. On confidentiality and conflicts, the opinion concludes that "a lawyer must not input any confidential information of a client into AI that lacks adequate confidentiality and security protections," because evolving large language models may carry information from one representation into another, implicating Rules 1.6, 1.7, and 1.9.

On communication, the opinion directs that under Rule 1.4 the lawyer inform the client of the benefits, risks, and limits of generative AI and, with the client, determine whether its use serves the client's objectives. On candor and meritorious claims, it concludes that "whether a baseless argument is made with the assistance of AI or not is irrelevant; the lawyer is responsible," and that under Rule 3.3 the lawyer must guard against false statements or evidence and take reasonable remedial measures. On supervision, it concludes the duties under Rules 5.1 and 5.3 that apply to paralegals and consultants apply to AI, with Rule 5.3 applying with equal force where AI functions like a human. On unauthorized practice, the opinion concludes that "there must always be a human element in the legal work product," so AI does not give legal advice or exercise legal judgment without a licensed attorney, and Rule 8.4(c) bars dishonest or deceitful use. As the opinion sums up, these tools "do not replace personal reviews of cases, statutes, and other legislative materials."

In practice

Under this opinion, a Pennsylvania lawyer using generative AI must understand the tool well enough to use it competently, independently verify every citation and factual output, keep confidential client information out of AI that lacks adequate protection, and disclose the benefits, risks, and limits to the client. The opinion holds the lawyer remains fully responsible for any baseless argument or false statement regardless of AI's involvement, must supervise AI as nonlawyer assistance under Rules 5.1 and 5.3, and must keep a human element in the work product so AI does not engage in the unauthorized practice of law.

Common questions

Q: Am I allowed to use ChatGPT or similar tools as a Pennsylvania lawyer?

A: Yes. The opinion concludes that, with appropriate safeguards, lawyers may use AI in accordance with the Rules of Professional Conduct.

Q: Do I have to check the cases and citations AI gives me?

A: Yes. The opinion concludes competence requires checking and verifying all citations and cited material, and notes lawyers have been sanctioned for filing nonexistent AI-generated cases.

Q: Can I paste client information into an AI tool?

A: The opinion concludes a lawyer "must not input any confidential information of a client into AI that lacks adequate confidentiality and security protections."

Q: Who is responsible if AI produces a bad argument?

A: The lawyer. The opinion concludes that whether a baseless argument is AI-assisted "is irrelevant; the lawyer is responsible."

Q: Can AI do the legal work on its own?

A: No. The opinion concludes "there must always be a human element in the legal work product," and AI must not give legal advice or exercise legal judgment without a licensed attorney.

Background and rules framework

The opinion interprets Pennsylvania Rules of Professional Conduct 1.1 (competence, and Comment [8] on technology), 1.4 (communication), 1.6 (confidentiality), 1.7 and 1.9 (conflicts), 3.1 (meritorious claims), 3.3 (candor toward the tribunal), 5.1 and 5.3 (supervision of lawyers and nonlawyer assistance), 5.5 (unauthorized practice), and 8.4(c) (dishonesty). These track the ABA Model Rules of the same numbers.

Citations and references

Rules of Professional Conduct:

• Pa.R.P.C. 1.1, 1.4, 1.6; ABA Model Rules 1.1, 1.4, 1.6
• Pa.R.P.C. 3.1, 3.3; ABA Model Rules 3.1, 3.3
• Pa.R.P.C. 5.1, 5.3, 5.5, 8.4(c); ABA Model Rules 5.1, 5.3, 5.5, 8.4

Other opinions cited:

• Florida State Bar Ethics Op. 24-1 (2024): lawyers may use generative AI
• State Bar of California COPRAC Practical Guidance (2023): generative AI risks
• PBA Formal Op. 2011-200: cloud computing confidentiality

Cases:

• Mata v. Avianca, Inc., 2023 WL 4114965 (S.D.N.Y. 2023), sanctions for fake AI cases
• Park v. Kim, 2024 WL 332478 (2d Cir. Jan. 30, 2024), fake case referred to grievance panel
• People v. Crabill, 23PDJ067 (Colo. Nov. 22, 2023), suspension for ChatGPT cases

See also

• ABA Formal Op. 512: Generative Artificial Intelligence Tools
• TX Ethics Op. 705: Generative AI in the Practice of Law
• FL Bar Ethics Op. 24-1: Generative AI

Source

• Landing page: PBA Ethics Opinions (Public)
• Original PDF: Joint Formal Opinion 2024-200.pdf