You Can Use AI in Your Law Practice. Your Ethical Duties Don't Change.

Plenty of attorneys still treat AI as a line they should not cross. The bar associations disagree. As of early 2026, more than 35 state bars have issued guidance, and the message is consistent: you are allowed to use AI, and your existing duties still apply.

The national framework is ABA Formal Opinion 512, issued in July 2024. It does not create new rules. It says the rules you already follow, competence, confidentiality, client communication, and reasonable fees, govern how you use AI. That is reassuring, because it means you already know the standard. You just have to apply it to a new tool.

Confidentiality is the duty most affected. Model Rule 1.6 requires reasonable efforts to prevent disclosure of client information. In practice, that means one bright line: never paste confidential client data into a free, public AI tool. Consumer chatbots may store what you type and use it to train future models. A firm-grade tool with a business agreement does not. Same technology, very different risk.

A few states have added specifics worth knowing. Florida says you must disclose AI use when it affects what a client is billed. Texas requires human review of anything AI produces, after several lawyers were sanctioned for filing fake case citations the AI invented. New York addressed AI tools that record and transcribe client meetings, stressing consent.

There is a competence angle too, and it cuts both ways. Opinion 512 ties AI to your duty of competence, which means you are expected to understand the tools well enough to use them responsibly. That includes knowing when AI gets things wrong. The well-publicized cases of lawyers filing briefs with invented citations were not failures of the technology. They were failures to review. The duty of competence does not require you to use AI, but if you do, it requires you to check it.

The practical takeaway: AI is not a compliance problem if you handle it like every other vendor. Check how the tool stores data, keep client details out of public tools, and review every output before it leaves your office. Lawyers who ignore AI entirely are not safer. They are just slower, and they are leaving the efficiency gains to the firm down the street.