AI for Lawyers in Small Practices: Safe Uses and Red Lines

By Alex Mercer · Last reviewed 2026-10-07

AI for small law firms can save real hours on drafting, summarizing and admin, but only if you treat it as an unsupervised junior who never touches confidential data without a vetted tool and whose every citation you check. That is roughly what the American Bar Association said in Formal Opinion 512 in July 2024. This guide is for solo and small-practice lawyers and their staff who want a short list of safe uses and clear red lines. It is general information, not legal or ethics advice; your state's rules control.

What the ethics guidance actually says

ABA Formal Opinion 512, released on July 29, 2024, was the ABA's first formal ethics opinion on generative AI. It does not create new rules. It applies existing Model Rules to AI tools:

Competence (Rule 1.1): understand the benefits and risks of the tool you use, at a reasonable level.
Confidentiality (Rule 1.6): make reasonable efforts to prevent disclosure of client information. Be especially careful with tools that may use your inputs to train the system, and read the opinion itself for what it says about client consent.
Communication (Rule 1.4): in some situations you must tell clients you are using AI.
Meritorious claims and candor (Rules 3.1, 3.3): check AI output, including citations, before you file it.
Supervision (Rules 5.1, 5.3): set firm policies and train staff who use AI.
Fees (Rule 1.5): fees must stay reasonable when AI speeds up the work; read the opinion's fee section before setting a billing approach.

State bars have issued their own opinions. Florida Bar Ethics Opinion 24-1 (January 2024), for example, says lawyers may use generative AI but must protect confidentiality, provide competent and accurate services, avoid improper billing such as double-billing, and follow advertising rules. Check whether your state bar has published similar guidance.

A step-by-step approach to AI for small law firms

Read your bar's guidance. Start with ABA Opinion 512 and any opinion from your state.
Sort your tasks into three buckets. Public or non-client work, client work with confidential data, and work that goes to a court or opposing counsel.
Vet each tool's data terms. Find out whether inputs are stored, for how long, and whether they train the model. Business and legal-specific plans often differ from consumer plans; confirm in writing.
Write a one-page firm AI policy. Approved tools, banned uses, who verifies output, and how to record AI use on a matter.
Update client communication. Check what the opinion and your state bar say about telling clients about AI use, and explain it in plain terms.
Review billing. Decide how you will bill AI-assisted work before the first invoice goes out.

Our explainer on whether it is safe to upload files to ChatGPT walks through the data-settings questions in step 3.

Safe uses: a practical task list

These are low-risk starting points for a small practice, assuming you keep confidential details out of tools you have not vetted:

Marketing and website copy about your practice areas, checked against advertising rules.
Plain-language explanations of general legal concepts for client newsletters.
Template first drafts of routine letters, with names and facts added afterward.
Summaries of public documents such as statutes, regulations or published opinions, which you then verify.
Admin work: meeting agendas, checklists, intake form wording and email tone edits.
Brainstorming issues or counterarguments, treated as a prompt for your own research, never as research.

Clio

Clio is a practice management platform that markets AI features. If you use any AI built into your practice management system, it works on client data you already store there, so ask the vendor which plans include it and how client data is handled before switching it on.

CoCounsel

CoCounsel is a Thomson Reuters AI product for legal work; check Thomson Reuters' own pages for what it currently covers. Research tools built on legal databases aim to ground answers in real sources, but you still verify every cited authority.

Lexis+ AI

Lexis+ AI is a LexisNexis AI product; check LexisNexis's own pages for its current features. The same rule applies: read the cited case yourself before relying on it.

We have not tested these tools and make no claims about their accuracy. Pricing for legal research products is usually quoted by sales, so ask for a written quote and data-processing terms.

Mistakes to avoid

ChatGPT

The common mistakes with general assistants like ChatGPT are pasting a client's confidential facts into a consumer account, and trusting a case citation it produced. General chatbots can generate citations that look real and are not. Opinion 512 expects you to check output before relying on it, and filing an unchecked brief risks a candor problem.

ABA guidance

Do not read the ABA opinion as a green light or a ban. It says the duties you already have apply to AI. Other mistakes include assuming a one-line AI clause in an engagement letter settles every consent question, billing without thinking through how AI changed the time spent, and letting staff use AI with no written policy. Your state bar's guidance controls, so check it.

When not to use AI for this

Keep AI away from these tasks, or use it only under close review:

Anything filed with a court that you have not verified line by line, including every citation and quotation.
Confidential client information in an unvetted tool, or in any tool where inputs may train the model, without informed client consent.
Final legal judgment. Strategy, settlement advice and whether to bring a claim stay with you.
Client-facing chatbots that give legal advice. These raise questions about unauthorized practice, advertising and confidentiality; get your bar's guidance first.

If a task needs both confidential facts and a public chatbot, it is not ready for AI. Strip the facts or use a vetted tool.

FAQ

Can lawyers use ChatGPT?

Yes, with care. ABA Formal Opinion 512 and Florida Bar Opinion 24-1 both accept that lawyers may use generative AI, provided they protect client confidentiality, verify the output and bill fairly. Do not paste confidential client information into a consumer tool without understanding its data terms.

What are the ethics rules for AI in law?

There is no single AI rule. ABA Formal Opinion 512 (July 2024) applies existing Model Rules on competence, confidentiality, communication, meritorious claims, candor to the tribunal, supervision and fees. Your state bar's rules and opinions control, so check them too.

What AI do small law firms use?

Common choices are general assistants such as ChatGPT for non-confidential drafting, AI built into practice management software such as Clio, and legal research tools such as Thomson Reuters CoCounsel and Lexis+ AI. Vet each one's confidentiality terms before use.

Conclusion

Spend one hour this week reading ABA Opinion 512 and your state bar's guidance, then write a one-page AI policy that names approved tools and the red lines above. Start with the safe task list, keep client data out until a tool's terms are vetted, and verify everything before it leaves the office. For how other professions are approaching the same questions, see our guide to AI tools by profession, and if you work with a remote assistant, our page on AI for virtual assistants covers delegation.

Sources and methodology

Last reviewed: 2026-10-07. Based on official ethics guidance and vendor pages; no tools were tested.

ABA Formal Opinion 512, Generative Artificial Intelligence Tools (July 29, 2024): https://www.americanbar.org/content/dam/aba/administrative/professional_responsibility/ethics-opinions/aba-formal-opinion-512.pdf
ABA news release on Opinion 512: https://www.americanbar.org/news/abanews/aba-news-archives/2024/07/aba-issues-first-ethics-guidance-ai-tools/
The Florida Bar, Ethics Opinion 24-1: https://www.floridabar.org/etopinions/opinion-24-1/