ABA Formal Opinion 512 explained: what every lawyer using AI must do

What Opinion 512 is, and what it is not

Formal Opinion 512 was issued by the ABA Standing Committee on Ethics and Professional Responsibility on July 29, 2024. It was the ABA’s first formal opinion on generative AI, and its message is permissive: lawyers may use these tools to serve clients, provided they meet the obligations they already have.

Two points frame everything else:

• It applies existing rules. The opinion does not create a special AI rulebook. It walks through the Model Rules on competence, confidentiality, communication, meritorious claims and candor, supervision and fees, and explains how each applies.
• It is guidance, not law. ABA opinions are persuasive. Your obligations come from the rules of the jurisdictions where you practise, and some of those jurisdictions have now gone further, including through court rules and, in California, a statute. (See our state-by-state guide.)

The opinion describes generative AI tools as producing a statistically probable output when prompted. That one sentence explains most of what follows: a tool that predicts likely text can be fluent and wrong at the same time.

Duty 1: Competence (Rule 1.1)

Competence now includes a reasonable understanding of the capabilities and limitations of the AI tools you use. You do not need to be a technologist, but you need to know enough to use a tool appropriately and to spot when it is likely to fail.

The opinion’s central practical requirement is verification. Lawyers may not rely on AI output without an appropriate degree of independent verification or review. How much checking is appropriate depends on the tool and the task: brainstorming ideas needs less scrutiny than drafting a brief or a contract clause that a client will rely on.

In practice:

• Learn how each tool you use works, including whether it is grounded in a legal database.
• Check every authority and quotation in anything that will be relied on, filed or sent.
• Keep learning. The tools change every few months.

Duty 2: Confidentiality (Rule 1.6)

This is the section most lawyers need to read twice. The opinion explains that the risk of disclosure depends on the client, the matter, the task and the specific tool.

It pays particular attention to self-learning tools, meaning tools that may use your inputs to improve. With those tools, confidential information you enter could later surface in an output to someone else. The opinion indicates that before inputting confidential client information into such a tool, a lawyer needs the client’s informed consent, and that boilerplate provisions in an engagement letter are not enough to provide it.

The opinion also expects lawyers to understand the terms of use, privacy policy and related contractual terms of the tools they use, or to consult someone who can, such as IT or cybersecurity staff.

In practice:

• Use business or enterprise plans whose terms prohibit training on your data. Avoid personal and free accounts for client work.
• Input only what the task requires, and remove identifying details where you can. Our free Confidentiality Redactor helps.
• Where a tool may learn from inputs, explain the specific risks to the client and get consent for that use.

Duty 3: Communication (Rule 1.4)

Opinion 512 does not require you to announce every use of AI. It identifies situations where disclosure is required:

1. The client asks how you are using AI.
2. Your engagement terms or the client’s instructions (for example, outside counsel guidelines) require it.
3. Informed consent is needed, as discussed above.
4. AI output will influence a significant decision in the representation.

Outside those situations, the opinion asks lawyers to consider the client’s needs and expectations, the scope of the representation and the sensitivity of the information. Many firms now choose to describe their AI use in engagement letters as a matter of trust. Our Client AI Disclosure Clause Generator drafts that language.

Duty 4: Meritorious claims and candor (Rules 3.1, 3.3 and 8.4(c))

Lawyers must not bring frivolous claims, make false statements of law or fact to a tribunal, or engage in misrepresentation. AI-assisted filings must be reviewed carefully so they contain no misstatements of law or fact and do not omit controlling authority.

This is the duty most often breached in practice, through fabricated citations. Courts have sanctioned lawyers in a large and growing number of cases, and in March 2026 the Sixth Circuit stated in Whiting v. City of Athens that no filing should contain citations the lawyer has not personally read and verified. See our sanctions tracker.

Duty 5: Supervision (Rules 5.1 and 5.3)

Lawyers with managerial authority must make reasonable efforts to ensure the firm has measures in place so that everyone complies with their professional obligations when using AI. The opinion points to clear policies on permissible use and to training. Supervisory duties also reach non-lawyers and third-party vendors, which is why vendor due diligence matters.

In practice:

• Adopt a written AI use policy (start with our AI Use Policy Generator).
• Train lawyers and staff on approved tools, prompting, confidentiality and verification.
• Review vendors’ terms, security and data handling before approval (use our Vendor Due Diligence Builder).

Duty 6: Fees (Rule 1.5)

The fee guidance is concrete:

• Hourly billing: bill for actual time. Time spent prompting, reviewing and verifying is billable work; time saved is not.
• Flat fees: a flat fee may become unreasonable if AI lets you finish the work far faster than the fee assumed.
• Learning: do not charge clients for time spent learning a tool you will use generally, unless the client asks you to use a particular tool.
• Costs: general-purpose AI built into your software is overhead. The cost of a third-party tool used for a specific matter may be charged as an expense where reasonable and agreed.

We cover this in depth in Should lawyers bill for AI time?

A one-page compliance checklist

Download the full version as a free PDF: the ABA Formal Opinion 512 Compliance Checklist.

How states and courts have built on it

Since 2024, many state bars have issued their own opinions or guidance, broadly consistent with Opinion 512. Some jurisdictions have gone further: Colorado amended its Rules of Professional Conduct in January 2026, New York’s courts adopted Part 161 effective June 1, 2026, and California enacted SB 574, effective January 1, 2027. Check where you practise with our State Bar AI Guidance Navigator.

Frequently asked questions

For your desk

AI for Lawyers 2026

The most complete edition yet: the tools worth your time, the regulation you cannot ignore, and a working method for using AI across a legal practice.

See it on Amazon

This article is general educational information, not legal advice. Rules differ by jurisdiction and change often; check the primary sources and your own professional rules before relying on it.