What the Latest State Bar AI Opinions Mean for How You Practice

The Trend Is Now a Standard

State bar AI ethics guidance started as a trickle in 2023. By mid-2024 it was a steady stream. In 2026 it is a chorus - and the message is consistent enough that any practicing attorney can be held to it, regardless of which jurisdiction they appear in.

This month brought another round of opinions across several states. None of them broke new ground in isolation. Together, they cement the standard. This article walks through what is consistent across the latest guidance, what it actually requires in daily practice, and the specific operational changes most firms should make this week.

For the broader framework these opinions build on, see AI ethics for lawyers: Rules 1.1, 1.6, and 5.3.

The Three Principles Showing Up in Every Opinion

Read three recent state bar AI opinions side by side and the same three principles appear, with slightly different language each time:

1. Competence requires AI literacy - even if you do not use it

The ABA's 2024 Formal Opinion 512 made this explicit. Every opinion since has reinforced it. The duty of competence under Rule 1.1 now expressly includes "understanding the benefits and risks" of generative AI in legal practice.

You do not have to use AI. You do have to understand it. "I did not know AI hallucinates" is not a defense, and the more recent opinions are explicit that this defense is exhausted as of 2026.

2. Confidentiality requires intentional tool selection

Rule 1.6 obligations are extending. The latest opinions are explicit that pasting client information into a consumer-tier AI tool that may train on user data can constitute disclosure - and that lawyers are responsible for understanding the data handling of the tools they use.

The practical implication: enterprise-tier subscriptions (ChatGPT Team or Enterprise, Claude Team or Enterprise, Gemini Workspace) are increasingly the floor for any AI work touching client information. We covered the comparison framework in ChatGPT vs. Claude vs. Gemini for lawyers.

3. Supervision of AI is a Rule 5.3 obligation

Multiple bars have now framed AI tools as analogous to non-lawyer assistants under Rule 5.3. The supervisor's obligation applies. Verification of AI output - especially case citations and statutory references - is not optional. See AI hallucinations in legal work for the verification workflow this should produce.

What Is Different in the Latest Opinions

Three operational themes are sharper in the most recent batch than they were in earlier guidance.

Disclosure to clients is no longer "recommended" - it is becoming standard

Earlier opinions framed client disclosure of AI use as "strongly recommended" or "best practice." The latest opinions are noticeably firmer. Several explicitly require disclosure when AI is used substantively in client work, and most now expect engagement letter language addressing AI use as a routine matter.

Action: update your engagement letter this week if you have not already. We published sample language in should you bill AI time to clients.

Billing rules around AI efficiency are getting concrete

A consistent theme: lawyers may not bill for the hours AI "saved." If AI does in 20 minutes what used to take 4 hours, the bill reflects 20 minutes. The latest opinions go further by specifically addressing the "technology fee" model some firms have adopted - it is generally acceptable if disclosed in the engagement letter and reasonable, but cannot be used as a vehicle to recover the hourly-billing shortfall.

The strategic implication: firms that shift more practice areas to flat or value-based billing are positioning more cleanly under the emerging consensus.

Court rules are being treated as primary obligations

Several opinions explicitly note that local court rules requiring AI disclosure in filings impose distinct obligations on top of bar requirements. Even where the bar does not require court disclosure, the court rule controls. See our follow-up article: Federal courts are mandating AI disclosure in filings.

What This Means Operationally This Week

If you are practicing in any U.S. jurisdiction, here is what the latest guidance suggests you should already have in place - and if you do not, what to do this week:

1. A written AI use policy for your firm. Even a one-page document is sufficient if it identifies approved tools, prohibited categories of input, and verification requirements.

1. Engagement letter language on AI use. A short paragraph disclosing that the firm uses AI tools, that all work is supervised by a licensed attorney, and that AI tool costs may be included in a technology fee.

1. A verification protocol for AI-assisted filings. Every case citation manually verified in a primary-source database. Every statute reference confirmed against current code. No exceptions.

1. Tier auditing of your AI tools. Anyone in your firm using consumer-tier AI for client work is creating exposure. Enterprise tiers are inexpensive and resolve the issue cleanly.

1. A training touchpoint at least annually. Attorneys need to demonstrate understanding of the tools and their limitations. Document that the training occurred. (Our AI Training program handles this for firms that want a structured rollout.)

The Risk of Doing Nothing

The cost of compliance with the emerging standard is low - a few hours of policy work, an engagement letter update, modest subscription upgrades, and a verification habit. The cost of non-compliance is meaningfully higher: by mid-2026, there have been dozens of published sanctions orders, fee disgorgement orders, and ethics referrals related to AI misuse in filings.

The pattern is consistent: the firms getting in trouble are not the ones using AI heavily. They are the ones using it without the operational guardrails the bars now expect. The work to put those guardrails in place is straightforward. Doing it before you need it is the entire point.

If you want help building this infrastructure into your firm - written policies, training, verification workflows, billing updates - our AI Operating System service handles the full implementation. Or book a 30-minute consult and we will walk through where your firm sits today.