The Attorney's Guide to AI Ethics: ABA Opinion 512 and California's Rules
9/5/20264 min read
The Attorney's Guide to AI Ethics: ABA Opinion 512 and California's Rules
If you want the whole picture of your ethical obligations when using generative AI — in one place — this is it. There is no single "AI rule." Instead, your existing duties of professional responsibility each apply to AI in a specific way. This guide walks through the full framework and links to a deeper treatment of each duty.
The good news, and the theme of everything below: AI didn't create new ethics rules. It sharpened the ones you already have.
This is general information, not legal advice.
The two anchors: ABA Opinion 512 and California's guidance
Two sources define the current landscape for California attorneys.
ABA Formal Opinion 512 (July 29, 2024) is the American Bar Association's first formal ethics guidance on generative AI. It doesn't invent a new regime — it applies six existing Model Rules to AI use. Nearly every state bar that has acted since, California included, builds on it as the baseline.
California's own framework is more directly binding on California lawyers. The State Bar's Practical Guidance for the Use of Generative Artificial Intelligence in the Practice of Law (first issued 2023, updated 2026) organizes AI duties around the same core obligations. And following an August 2025 directive from the California Supreme Court, COPRAC approved a package of proposed amendments — weaving AI-specific language into Rules 1.1, 1.4, 1.6, 3.3, 5.1, and 5.3 — at its March 13, 2026 meeting, with a public comment period that closed in early May 2026 and updated Practical Guidance approved by the Board of Trustees in May 2026. Those rule amendments are proposed, not yet adopted, and remain subject to the California Supreme Court's authority; the underlying duties, however, are already in force.
Remember the distinction throughout: ABA opinions are persuasive guidance interpreting the Model Rules. Your binding authority is California's Rules of Professional Conduct and the Business and Professions Code.
The six duties, and how AI touches each
1. Competence (Rule 1.1).
You must understand a tool's capabilities, limitations, and risks before using it in client work — and independently verify its output. This is the foundation for everything else, and it's the primary anchor for autonomous, "agentic" tools. → See: Supervising Agentic AI.
2. Confidentiality (Rule 1.6; § 6068(e)).
Protect client information. Entering confidential or identifying data into a public AI tool can itself be a disclosure if the tool may retain, train on, or share your inputs. Vet the tool before the data goes in. → See: Attorney Confidentiality and Public AI Tools; AI Vendor Due Diligence.
3. Communication (Rule 1.4).
Keep clients reasonably informed. There's no blanket duty to announce every use of AI, but disclosure becomes appropriate where AI materially affects the scope, cost, or decision-making of a representation. → See: Client Communication and AI Disclosure.
4. Candor toward the tribunal (Rule 3.3).
Never file what you haven't verified. AI-hallucinated citations have already drawn published sanctions in California — the responsible attorney must personally read and verify every citation in a filing. → See: What Noland Means for Your Filing Workflow.
5. Supervision (Rules 5.1 and 5.3).
Managerial lawyers must establish firm policies governing AI use, and all lawyers must supervise the staff who use these tools. A written AI use policy is how you make that duty demonstrable. → See: Building a Law-Firm AI Use Policy.
6. Reasonable (and non-unconscionable) fees (Rule 1.5).
Bill for time actually spent, not the pre-AI estimate; don't charge clients to learn the tool; let efficiency gains benefit the client. Note California's stricter framing: its Rule 1.5 prohibits unconscionable fees, not merely unreasonable ones. → See: Billing for AI Work.
How to use this framework
You don't need to memorize six rules before opening a chatbot. In practice, three habits cover most of the risk: vet the tool before client data touches it, verify the output before it's filed or sent, and document your decisions — the policy, the vendor review, the citation check. Those three moves satisfy the spirit of all six duties, and they're the difference between the lawyers who use AI well and the ones who end up in a published opinion.
Frequently asked questions
Is there a single ethics rule for lawyers using AI?
No. Your existing duties — competence, confidentiality, communication, candor, supervision, and fees — each apply to AI. ABA Opinion 512 and California's guidance map AI use onto those existing rules.
Are California's AI rule amendments in effect?
Not yet. As of 2026 they are proposed amendments still in the rulemaking process, subject to the California Supreme Court. The underlying duties apply now regardless.
Does ABA Opinion 512 bind California attorneys?
Not directly — it's persuasive guidance interpreting the ABA Model Rules. California's own Rules of Professional Conduct and State Bar guidance are your governing authority, though they cover the same core duties.
What are the three most important habits?
Vet the tool before inputting client data, verify AI output before filing or sending it, and document your policies and reviews.
This article is general information, not legal advice. Confirm the current State Bar guidance and any applicable rule before relying on specifics.
Sources
California Rules of Professional Conduct 1.1, 1.4, 1.5, 1.6, 3.3, 5.1, 5.3
California Business & Professions Code § 6068(e)
ABA Formal Opinion 512 (July 29, 2024) — Generative Artificial Intelligence Tools
State Bar of California, Practical Guidance for the Use of Generative Artificial Intelligence in the Practice of Law (2023; updated 2026), and proposed amendments to the Rules of Professional Conduct
California Supreme Court directive to the State Bar (August 2025)