Can Lawyers Use ChatGPT? What the Ethics Rules Say

Can lawyers use ChatGPT? Yes: no ABA rule or state bar opinion we have read bans lawyers from using ChatGPT or other generative AI tools. What the guidance does say is that the existing duties of competence, confidentiality, candor and reasonable fees apply to them, and those duties shape which version of the tool a lawyer uses and what goes into it.

The main reference point is ABA Formal Opinion 512, issued July 29, 2024. It creates no new rule. It explains how the Model Rules already apply to generative AI, and state bars have since issued their own opinions that mostly reach the same place. Our explainer on Opinion 512 and our state bar opinion tracker cover both in detail. This piece describes what that guidance says. It is not legal advice, and each lawyer’s obligations depend on the rules of their own jurisdiction.

What the ABA and State Bars Say About ChatGPT

Opinion 512 starts with competence under Model Rule 1.1. It says lawyers do not need to be technology experts but do need a reasonable understanding of what a tool can and cannot do. It cites a Stanford study that found leading legal research tools hallucinated between 17% and 33% of the time, and it describes the level of checking a lawyer applies as depending on the tool and the task.

State bars have added detail. Florida’s Opinion 24-1, from January 19, 2024, treats confidentiality as the central risk and recommends getting a client’s informed consent before sharing confidential information with a third-party AI tool. North Carolina’s 2024 Formal Ethics Opinion 1, adopted November 1, 2024, says lawyers should avoid putting client-specific information into publicly available AI tools. California’s State Bar keeps a running list of other jurisdictions’ opinions on its AI resources page.

Why the Version of ChatGPT Matters

Much of the confidentiality analysis turns on what happens to the data a lawyer types in. Opinion 512 describes tools that learn from user input, which it calls self-learning tools, and says a client’s informed consent is required before information relating to the representation goes into one. It adds that general boilerplate in an engagement letter is not enough.

OpenAI draws a line between its consumer and business products. Its enterprise privacy page, updated January 8, 2026, says that by default it does not train its models on business data from ChatGPT Business, ChatGPT Enterprise and its API platform, and that those products have completed SOC 2 Type 2 audits. The same page says OpenAI does use data from versions of ChatGPT for individuals when training models. ChatGPT’s pricing page says Business plans start at two users.

That distinction is the kind of detail Opinion 512 suggests reading in a tool’s terms of use and privacy policy. Oregon’s Formal Opinion 2025-205 frames it as open versus closed models and describes informed consent as required before client information goes into an open one. Our glossary and AI for lawyers guide explain the terms.

Where Lawyers Using ChatGPT Have Gone Wrong

The best-known case is Mata v. Avianca. On June 22, 2023, Judge P. Kevin Castel of the Southern District of New York sanctioned two lawyers and their firm after a filing cited judicial opinions that ChatGPT had made up. The court imposed a $5,000 penalty and ordered letters to the client and to each judge falsely named as an author. The opinion makes clear that the problem was not using the tool but failing to check what it produced and then failing to come clean.

Courts have kept issuing sanctions since. We covered two recent federal orders in Two Federal Judges Put a Price on Unchecked AI Citations, and some courts now require lawyers to disclose AI use in filings, which our court disclosure tracker follows.

Billing is the other recurring issue. Opinion 512 says hourly billing can reflect only time actually spent, and North Carolina’s opinion gives the example of a lawyer who cannot bill three hours for work that took one.

Common Questions

Is it unethical for a lawyer to use ChatGPT?

No bar opinion we have read calls it unethical in itself. The opinions describe how existing duties of competence, confidentiality, supervision and candor apply to its use.

Can lawyers put client information into ChatGPT?

Opinion 512 says informed consent is required before client information goes into a tool that learns from user input. Florida recommends consent for third-party tools, and North Carolina says to avoid client-specific information in public tools. Whether a given ChatGPT plan trains on inputs is set out in OpenAI’s own terms.

Do lawyers have to verify what ChatGPT writes?

Opinion 512 ties the level of review to the tool and the task, and the sanctions cases show courts holding lawyers responsible for every citation they file, whatever produced it.

Can a lawyer bill for time saved by ChatGPT?

Opinion 512 and several state opinions say hourly bills can only reflect time actually spent. Virginia’s LEO 1901 takes a more permissive view of flat fees.

The Verdict

Real. Lawyers can and do use ChatGPT, and the ethics guidance is permissive. The practical questions are which plan, what data goes into it, and who checks the output, and those are the questions the bar opinions keep returning to.