ABA Formal Opinion 512, issued in July 2024, is the clearest statement the American Bar Association has yet made about generative AI in legal practice. It is not aspirational guidance. It is a direct application of existing Model Rules to tools your attorneys are likely already using.
What Opinion 512 Actually Says
The Opinion addresses four areas: competence (Rule 1.1), confidentiality (Rule 1.6), supervision (Rules 5.1 and 5.3), and candor (Rule 3.3). On each one, the message is the same: AI does not create new ethical rules, but it creates new ways to violate existing ones — and attorneys are responsible for understanding the difference.
On competence, the Opinion states that attorneys must understand the benefits and risks of AI tools relevant to their practice. This is not a suggestion. It is a professional obligation. Attorneys who use AI without understanding hallucination risk, training data limitations, or output verification requirements are not competent users of that tool.
On confidentiality, the Opinion flags the data retention practices of AI vendors as a specific concern. Uploading client information to an AI platform without reviewing how that platform stores, uses, and retains data may constitute unauthorized disclosure under Rule 1.6. The fact that an attorney didn't realize their data was being retained is not a defense.
On supervision, the Opinion makes clear that AI output is supervised work product — not autonomous work. Partners who allow associates or staff to use AI tools without documented oversight protocols are assuming personal disciplinary exposure for every unsupervised output that reaches a client or a court.
On candor, the Opinion is blunt: submitting AI-generated content — including citations, case summaries, or legal arguments — without independent verification is a violation of Rule 3.3. This is the rule most frequently cited in sanctions orders to date.
Why This Matters for Small Firms
Large firms have compliance departments, general counsel, and the budget to implement governance frameworks before regulators ask the question. Small firms typically do not.
The attorney who received a $5,000 sanctions order in federal court in 2024 for submitting fabricated AI-generated citations was a solo practitioner. The partner who was referred to the state bar for unauthorized client data disclosure was at a three-attorney firm. Small practices are not insulated from enforcement — they are often the most exposed because they lack the institutional infrastructure to catch these failures before they reach a judge.
The Governance Gap
Opinion 512 does not tell attorneys to stop using AI. It tells them to use it defensibly. That means documented policies, defined supervision protocols, explicit tool approvals, and a written record showing that the firm understood and managed its AI-related professional responsibility obligations.
Most small firms currently have none of these. That is the governance gap that creates direct disciplinary exposure.
What a Defensible Posture Looks Like
A defensible AI governance posture has five components:
1. A written AI use policy that identifies approved tools, defines acceptable use cases, and documents the firm's approach to output verification.
2. Confidentiality safeguards that address vendor data retention, prohibit client data from entering unapproved systems, and establish informed consent procedures where relevant.
3. Supervision protocols that define how AI-generated work product is reviewed before it reaches a client or court, and who is responsible for that review.
4. Documentation requirements that create an audit trail showing the firm's AI governance posture at any given point in time.
5. A periodic assessment cadence that updates governance as tools, regulations, and case law evolve.
This is exactly what the AI Governance Phase 0 Assessment measures — and what the KeaneAdvisors.AI engagement model is designed to build.
The question is not whether your firm needs AI governance. After Opinion 512, that question is settled. The question is whether you address it before a court, a client, or a bar authority asks you to explain why you didn't.