ABA Makes AI Competence Mandatory for Every US Lawyer — 18-Month Clock Starts Today

The American Bar Association's House of Delegates voted this morning to amend Model Rule of Professional Conduct 1.1 — making generative-AI competence an explicit, non-waivable duty for every lawyer in the United States. The 18-month state adoption clock starts today.

At 10:42 ET on May 16, 2026, the ABA House of Delegates passed Resolution 604 by a margin of 412–58 — formally amending Model Rule of Professional Conduct 1.1 to add Comment [9], a one-paragraph addition that quietly rewires the duty of competence for every lawyer in America. The new comment states that competent representation requires lawyers to understand the benefits, limitations, and risks of generative artificial intelligence tools relevant to their practice, and to maintain that understanding through ongoing education. It is the single most consequential ethics-rule change the ABA has shipped since the technology competence amendment of 2012.

What Comment [9] Actually Says

The new comment is short by design. It does not require lawyers to use AI. It does not endorse any specific tool. What it does, with surgical precision, is establish that ignorance of generative AI — its capabilities, its hallucination risk, its confidentiality posture, its bias profile, and its regulatory status — is now an ethics violation in any practice area where AI tools are reasonably available. The drafting committee borrowed the structural pattern of the 2012 Comment [8] technology amendment but tightened the language: where the 2012 rule used the phrase should keep abreast, the 2026 amendment uses shall maintain.

The Six Specific Competencies Now Required

• Capability awareness — understanding what current generative-AI tools can and cannot reliably do in your practice area.
• Hallucination risk — knowing that AI can fabricate citations, statutes, and facts, and the verification workflow that mitigates it.
• Confidentiality posture — understanding which tools train on inputs, which offer zero-retention enterprise tiers, and which violate Model Rule 1.6 outright.
• Supervisory duty — Model Rules 5.1 and 5.3 now extend to AI outputs the same way they extend to associate and paralegal work product.
• Disclosure judgment — knowing when AI use must be disclosed to clients, courts, or opposing counsel under applicable state rules.
• Billing integrity — understanding that billing a client for AI-generated work as if it were human attorney hours implicates Rules 1.5 and 8.4.

Why Now — and Why It Passed By 412–58

The vote was telegraphed for months but the margin was not. Three forces converged. First, the California Rule 2.150 disclosure mandate adopted on May 13 created an immediate national reference point — delegates who were on the fence saw a state act before the ABA did and decided the profession could not be seen lagging its largest bar. Second, the National Discipline Data Bank reported 287 documented AI-hallucination sanctions cases against US attorneys in calendar 2025, up from 41 the prior year, with malpractice carriers privately briefing delegates that the trend line was unsustainable. Third, the Thomson Reuters–Harvey deal announced May 14 made it impossible for any delegate to argue that AI was still a niche tool only used by Big Law.

The 18-Month State Adoption Clock

Model rules are not self-executing — each state supreme court must adopt the amendment for it to bind lawyers in that jurisdiction. The ABA's accompanying transition report sets a recommended adoption window of 18 months, putting the realistic deadline at November 2027. Six states — California, New York, Illinois, Texas, Florida, and Washington — have AI working groups already drafting state-specific adoption language; California and New York are expected to act before Labor Day. The remaining 44 jurisdictions will follow on the historical pattern of the 2012 technology amendment, which took an average of 4.3 years to reach universal adoption — a pace this amendment will likely beat by half.

What This Means for CLE Providers

The amendment effectively creates a mandatory continuing-legal-education category overnight. Within four hours of the vote, the four largest CLE providers — Practising Law Institute, ALI CLE, Lawline, and the ABA's own CLE arm — had published new Generative AI Competence certificate programs ranging from 4 to 16 credit hours. Expect state bars to begin requiring an annual AI-competence credit within the next 24 months; New York and California will lead. Firms that previously treated AI training as optional will need to treat it as mandatory compliance spend, on the same footing as anti-money-laundering and conflicts training.

Not knowing how generative AI works is no longer a professional preference. It is, as of today, a discipline-able failure of competence.

Five Things Every Firm Must Do This Quarter

• Inventory your AI usage. Every tool — sanctioned and shadow — used by any attorney, paralegal, or staff member. You cannot supervise what you have not inventoried.
• Publish an internal AI-use policy. Approved tools, confidentiality tier requirements, verification workflow, disclosure protocols. The policy is the first artifact bar counsel will request in any future ethics inquiry.
• Schedule firm-wide training. Every lawyer, not just the litigation department. Document attendance for the personnel file — it is your first line of defence under Rules 5.1 and 5.3.
• Update engagement letters. Add an AI-use clause explaining what tools the firm uses, how confidentiality is preserved, and how AI-assisted work is billed. Clients will start asking by Q3.
• Brief your malpractice carrier. Several major carriers are already drafting AI-specific exclusions; understand what your policy covers before a hallucination shows up in a filed brief.

What It Means for In-House Counsel

Comment [9] applies with equal force to in-house lawyers. General counsel teams that have rolled out enterprise AI tools without a written competence-and-training program are now exposed under the same rule that binds law-firm partners. The amendment also creates a meaningful procurement-leverage shift: in-house teams can now require outside counsel to certify AI competence and tool inventory as part of outside-counsel guidelines, and many Fortune-500 legal-operations groups are expected to add that requirement to their 2027 RFP cycles.

What It Means for Solo and Small-Firm Practitioners

The amendment hits hardest at the bottom of the market, where AI adoption has been ad-hoc and training budgets are thin. The ABA simultaneously announced a free AI Competence Starter CLE program for solo and small-firm members, available beginning June 1. Solos who skip it are betting that their state will be slow to adopt — a bet that historically has not paid off.

The Bottom Line

The era of debating whether lawyers should learn generative AI is officially over. Comment [9] settles it: every lawyer in the United States now has an affirmative, ongoing ethics duty to understand the AI tools relevant to their practice. The 18-month state-adoption clock starts today. The firms and in-house teams that treat this as a compliance project — not a technology project — will be the ones still standing in 2028.