AI for Law Firms: Use Cases, Risks and Ethics Rules

Every managing partner is asking the same question: how do we get the speed of AI without becoming the next firm in the news for citing cases that do not exist? The answer is not to ban the tools. It is to match each use case to the professional duties that already govern your practice.

This guide walks through how law firms use AI for research, drafting and intake, which ethics rules apply to each, and a practical rollout plan for small and mid-sized firms. It is written for US attorneys and firm administrators. This is not legal advice; check your own state’s rules and ethics opinions.

Where law firms are using AI today

AI in legal practice falls into 5 broad groups:

• Legal research. AI-assisted search in legal research platforms, plus summaries of cases, statutes and regulations.
• Drafting. First drafts of memos, letters, discovery requests, contract clauses and marketing content.
• Document review and analysis. E-discovery classification, contract review, due diligence and deposition summaries.
• Client intake and service. Website chat, intake forms that triage matters, appointment booking and status updates.
• Practice management. Time entry suggestions, billing narratives, calendaring and knowledge management.

The firms getting the most value pick 2 or 3 of these, choose tools built for legal work with proper data protections, and write clear rules about verification. For the wider technology picture, see how technology is changing law firms.

The ethics framework: ABA Formal Opinion 512

On July 29, 2024, the ABA Standing Committee on Ethics and Professional Responsibility issued Formal Opinion 512, its first formal opinion on generative AI. It does not create new rules. It explains how the existing Model Rules of Professional Conduct apply:

• Competence (Rule 1.1). Lawyers need a reasonable understanding of the capabilities and limitations of the specific AI tools they use. You do not need to be an expert, but you must know enough to use them responsibly, and you must check the output.
• Confidentiality (Rule 1.6). Before entering client information into a tool, evaluate the risk that it could be disclosed or used by others. For self-learning tools that may use inputs in ways that could reveal client information, the opinion says informed client consent is required, and boilerplate language in an engagement letter is not enough.
• Communication (Rule 1.4). In some situations you must tell clients about your AI use, for example when it affects the representation or the fee, or when the client asks.
• Meritorious claims and candor (Rules 3.1, 3.3). You are responsible for every citation and factual claim in a filing, whoever or whatever drafted it.
• Supervision (Rules 5.1, 5.3). Partners and supervising lawyers must set policies and train lawyers and staff on approved AI use, and vet outside vendors.
• Fees (Rule 1.5). Hourly billing must reflect time actually spent. You cannot bill a client for 3 hours of drafting that took 20 minutes with AI, or for time spent learning a general-purpose tool. Some tool costs may be passed through as expenses, while general software is usually overhead.

State bars have issued their own guidance too. The Florida Bar’s Advisory Opinion 24-1, adopted in January 2024, covers confidentiality, oversight, fees and advertising, and California, New York, New Jersey and others have issued guidance or reports. Check your state’s version, because the details differ.

Key takeaway: AI changes how legal work gets done, not who is responsible for it. The lawyer who signs the filing or sends the advice owns every word.

Use cases mapped to duties

Legal research and drafting: verification is non-negotiable

Generative AI can produce text that looks like a real case citation but is not. The best-known example is Mata v. Avianca in the Southern District of New York, where a court sanctioned lawyers and their firm in June 2023, imposing a $5,000 penalty after a brief cited fake cases generated by ChatGPT. The court noted that using a reliable AI tool is not improper in itself; the problem was failing to check the output and then standing by it. Courts have continued to sanction filings with fabricated citations since then, and many judges now have standing orders on AI use in filings.

Verification rules that work:

• Pull and read every case, statute and regulation cited, from an authoritative source.
• Check that quotes appear in the source and say what the draft claims.
• Run every citation through your citator.
• Prefer legal research platforms whose AI features ground answers in their own databases and link to sources, but still verify.
• Check local rules and judges’ standing orders for AI disclosure or certification requirements.

Our explainer on AI hallucinations covers why models fabricate and how grounding reduces it.

Client intake and chatbots

AI intake tools can answer basic questions, collect contact details and matter facts, book consultations and route urgent matters. Done well, they mean fewer missed leads after hours. The risks are specific to legal practice:

• Unintended attorney-client relationships. Visitors may believe they are getting legal advice. Use clear notices that the chatbot does not provide legal advice and that no relationship exists until an engagement is signed.
• Confidential information from prospects. Prospective clients have confidentiality protections under Rule 1.18. Limit what the bot asks for until a conflict check is done, and protect what it collects.
• Conflicts. Do not let the bot collect detailed facts before conflicts are cleared.
• Advertising rules. Bot responses are communications about your services and must not be false or misleading.

If you are weighing a bot versus a copilot for staff, see AI agents vs chatbots vs copilots.

Protecting client confidentiality

Confidentiality is where most firm AI policies start, and for good reason. Before approving any tool, ask the vendor:

1. Is our data used to train or improve models? Can we turn that off by contract, not just by setting?
2. Where is data stored and processed, and for how long?
3. Who at the vendor can access prompts and outputs?
4. What security certifications or audits does the vendor have?
5. Can we delete our data, and how is deletion confirmed?
6. Does the tool support matter-level access controls and ethical walls?

Consumer accounts of general chatbots are usually the wrong place for client data. Our guide on whether it is safe to put company data into AI has a vendor checklist that applies to firms too.

Where to start: a 6-step rollout for a small or mid-sized firm

1. Inventory what is already in use. Ask lawyers and staff which AI tools they use today, including personal accounts.
2. Write a firm AI policy. Cover approved tools, prohibited inputs, verification steps, client disclosure and billing. Our template for an AI acceptable use policy is a good starting outline.
3. Choose 2 low-risk pilots. Good starters are internal knowledge search, deposition summaries, first drafts of routine correspondence or marketing content.
4. Train and supervise. Short training on prompting, verification and confidentiality for every user, plus a named partner who owns the program.
5. Update engagement letters and billing practices. Explain AI use where relevant, and decide how AI tool costs are treated.
6. Review quarterly. Check errors caught, time saved, client feedback and any new ethics opinions or court orders in your jurisdictions.

For budgeting, see our breakdown of how much AI implementation costs, and for ownership questions about AI-drafted materials, read who owns AI-generated content.

Conclusion

AI for law firms is now a practical tool for research, drafting, review and intake, and the ethics rules are clear enough to use it responsibly. Know your tools, protect client confidences, verify every output, supervise everyone who uses them and bill honestly. Firms that build those habits now will capture the speed without the sanctions.

Sources

• ABA Formal Opinion 512: Generative Artificial Intelligence Tools (July 29, 2024) — competence, confidentiality, communication, candor, supervision and fee duties
• The Florida Bar: Ethics Opinion 24-1 — state guidance on generative AI covering confidentiality, oversight, fees and advertising
• ABA Model Rules of Professional Conduct — Rules 1.1, 1.4, 1.5, 1.6, 1.18, 3.3, 5.1, 5.3 and 7.1
• Mata v. Avianca, Inc., No. 22-cv-1461 (S.D.N.Y. June 22, 2023), Opinion and Order on Sanctions — sanctions for fabricated AI-generated citations