In July 2024, the ABA Standing Committee on Ethics and Professional Responsibility issued Formal Opinion 512, its first formal guidance on lawyers' use of generative artificial intelligence tools. For managing partners and administrators, the opinion is useful less as a rulebook than as a map. It identifies the existing Model Rules that AI use touches and signals how the ABA thinks about each.
This post summarizes the opinion in general terms, organized by rule, and turns each into a practical step. It is not legal advice. ABA opinions are persuasive rather than binding, and each state sets its own rules, so confirm with your state bar and ethics counsel.
Competence: Model Rule 1.1
The opinion discusses the duty of competence in relation to AI. In general terms, lawyers should have a reasonable understanding of the capabilities and limitations of the tools they use, and should not rely on output without appropriate review. The opinion notes that the degree of independent verification depends on the task and the tool.
What to do: Require training before access, treat AI output as a draft and require verification of every citation and factual claim before it leaves the firm.
Confidentiality: Model Rule 1.6
The opinion discusses the risk that information entered into a generative AI tool may be disclosed to others or used to train a model in ways that could affect client confidentiality. It notes that lawyers should evaluate the risks, including the tool's data practices, and that in some cases informed client consent may be required before inputting client information.
What to do: Approve specific tools after reviewing their terms, and prohibit client data in unapproved ones. Understand the difference between consumer and enterprise plans.
Communication: Model Rule 1.4
The opinion discusses when lawyers may need to tell clients about their use of AI. In general terms, communication may be needed when the client asks, when the use is significant to the representation or when a client's instructions or the engagement terms address it.
What to do: Add AI language to engagement letters, and check client outside counsel guidelines, some of which restrict or require notice of AI use.
Meritorious claims and candor: Model Rules 3.1 and 3.3
The opinion points out that lawyers remain responsible for the accuracy of what they file. Generative AI can produce plausible but incorrect statements, including nonexistent citations, and submitting them can raise candor and sanctions issues.
What to do: Make verification a documented step for anything filed with a tribunal. Some courts have standing orders on AI use, so have someone check the rules for each court you appear before.
Supervision: Model Rules 5.1 and 5.3
The opinion addresses managerial and supervisory responsibilities. In broad terms, firm leaders should establish clear policies on permissible AI use and should make reasonable efforts to ensure lawyers and nonlawyers comply, including through training.
What to do: Adopt a written policy, train everyone, and review how tools are used. Include vendors and outside providers in your oversight.
Fees: Model Rule 1.5
The opinion discusses billing in connection with AI. In general terms, fees must be reasonable, and lawyers billing hourly should bill for actual time spent. The opinion also discusses that general overhead costs for tools are typically not separately billed unless agreed with the client.
What to do: Decide how your firm will treat time saved by AI and any tool costs, and put it in engagement letters.
A practical action list for firm leaders
- Inventory current use. Ask what people are using now.
- Approve a short list of tools after vendor review, and say which data may go where.
- Write a short policy, one to three pages.
- Train attorneys and staff, with real examples.
- Update engagement letters to address AI use and billing.
- Check court rules and client guidelines for restrictions or disclosure requirements.
- Document verification for filed work.
- Review at least twice a year, since tools and guidance change quickly.
What the opinion does not do
It does not ban AI. It does not provide a checklist of approved tools. It does not settle every question, and it expressly recognizes that the technology and the guidance will continue to evolve. State bars have issued their own AI guidance, with differences in emphasis, so local guidance matters.
A note on tone
Opinion 512 reads as neither alarmist nor dismissive. It treats generative AI as a tool whose use must fit within long-standing duties. That is a good frame for firm leadership: avoid both blanket bans, which drive use underground, and uncontrolled adoption, which creates risk without a record.
How Counsel Cyber can help
Counsel Cyber helps firms apply guidance like Opinion 512 by reviewing AI vendors, configuring approved tools securely and training staff. If you would like help turning the action list into a plan, we are glad to talk it through.