Bar Association Guidance on AI: A 2026 State-by-State Map

Bar Association Guidance on AI: A 2026 State-by-State Map

If you are a lawyer using AI in your practice right now, or advising a company that does, you are navigating a regulatory landscape that looks less like a coherent framework and more like a quilt stitched together by 50 different people who never talked to each other. State bar associations have been issuing guidance on AI use at an accelerating pace since late 2023, and by mid-2026, we have enough data points to start drawing a real map. The problem is that the map is messy.

Here is where things stand in the jurisdictions that matter most, and what the broader patchwork looks like for firms operating across state lines.

California: Detailed and Prescriptive

California, predictably, moved early and went deep. The State Bar of California issued Formal Opinion 2024-01 in November 2024, which addressed generative AI use across multiple dimensions: competence under Rule 1.1, confidentiality under Rule 1.6, and supervision obligations under Rules 5.1 and 5.3. The opinion explicitly stated that lawyers have a duty to understand the technology "sufficiently to comply with these obligations," which is a higher bar than simply knowing AI exists.

In March 2026, the California State Bar's Standing Committee on Professional Responsibility followed up with Practical Guidance Bulletin 2026-03, which added specificity around client disclosure. The bulletin recommends, though does not yet require, that attorneys inform clients when AI tools are used in substantive legal work. It also flags billing practices, cautioning that billing AI-assisted research at the same rate as traditional research may raise issues under Rule 1.5's fee reasonableness standard.

California's approach is notable because it treats AI not as a novelty but as a routine tool that falls squarely within existing ethical infrastructure. No new rules, just rigorous application of old ones.

Florida: The Disclosure Mandate

Florida took a different path and arguably a bolder one. The Florida Bar's Board of Governors approved amendments to the Rules Regulating the Florida Bar in January 2025 that created an affirmative disclosure obligation. Under amended Rule 4-1.6, attorneys must inform clients in writing when AI tools are used in any matter where the AI processes client-confidential information. This went into effect on July 1, 2025.

The Florida Bar also issued Ethics Opinion 24-1 in late 2024, which addressed the Mata v. Avianca problem head-on. The opinion states that submitting AI-generated content to a tribunal without independent verification constitutes a violation of Rule 4-3.3 (candor toward the tribunal) and potentially Rule 4-8.4 (misconduct). Florida has already seen at least two public reprimands tied to AI-generated filings with fabricated citations, one in the Southern District in September 2025 and another in a state court matter in Hillsborough County in early 2026.

For firms operating in Florida, the written disclosure requirement is the headline. It creates a paper trail, and that paper trail creates liability exposure if the disclosure is missing or incomplete.

New York: Slow but Influential

New York has been surprisingly measured. The New York State Bar Association's Task Force on Artificial Intelligence released its final report in April 2025, running over 100 pages, but the NYSBA stopped short of recommending rule changes. Instead, the report urged practitioners to treat AI outputs as "preliminary drafts requiring the same scrutiny as work product from a junior associate." That framing has been widely cited in other jurisdictions.

The New York City Bar Association was more aggressive, issuing Formal Opinion 2025-2 in August 2025, which specifically addressed confidentiality risks when using cloud-based AI tools. The opinion requires attorneys to conduct due diligence on the AI provider's data handling practices before inputting any client information. It references the ABA's Model Rule 1.6(c) and its requirement to make "reasonable efforts" to prevent unauthorized disclosure.

Several New York federal courts have also imposed their own standing orders. The Southern District of New York's amended General Order 2024-03, updated in February 2026, requires attorneys to certify whether AI was used in the preparation of any filing and to identify the specific tools used. The Eastern District adopted a nearly identical order in May 2026.

Texas: Principles Over Prescriptions

Texas has taken the lightest touch among major jurisdictions. The State Bar of Texas issued Proposed Ethics Opinion No. 708 in June 2025, which was adopted in final form in December 2025. The opinion addresses AI use primarily through the lens of competence (Rule 1.01) and confidentiality (Rule 1.05), but it avoids specific mandates around disclosure or billing.

The Texas opinion is notable for what it does not say. There is no recommended disclosure framework, no specific guidance on billing AI-assisted work, and no mention of court filing certifications. The opinion essentially tells lawyers to use good judgment and comply with existing rules. For a state with over 100,000 active attorneys, that leaves a lot of room for interpretation.

The Texas approach may reflect a philosophical preference for flexibility, but it also creates uncertainty. Lawyers in Texas who want to adopt AI tools aggressively have less cover if something goes wrong, because there is no detailed guidance to point to as evidence of good-faith compliance.

The Broader Patchwork

Beyond the big four, the landscape is fragmented. As of June 2026, at least 34 state bar associations have issued some form of guidance on AI, ranging from full formal opinions to brief advisory notices. A few highlights worth tracking:

• Illinois: The ISBA issued Advisory Opinion 25-01 in March 2025, focusing heavily on supervision obligations when non-lawyer staff use AI tools. This is one of the few opinions that directly addresses the paralegal and legal assistant use case.
• New Jersey: The Advisory Committee on Professional Ethics published Opinion 742 in October 2025, which aligns closely with Florida's disclosure approach but stops short of making it mandatory.
• Colorado: Adopted a formal amendment to Rule 1.0 in January 2026, adding a definition of "artificial intelligence tool" to the state's Rules of Professional Conduct. Colorado is the first state to embed AI terminology directly into its ethics rules.
• Georgia: Issued a brief advisory in September 2025 that essentially restates existing competence and confidentiality obligations without AI-specific guidance. Minimal practical impact.
• Washington: The WSBA's AI Task Force released interim recommendations in April 2026 that propose a tiered disclosure system based on the risk level of the AI application. This is the most nuanced framework proposed by any state bar to date, though it remains in draft form.

The ABA's Formal Opinion 512, issued in July 2025, provides a national baseline but lacks enforcement power. It is useful as a reference point, particularly its treatment of the duty to supervise AI outputs under Model Rules 5.1 and 5.3, but state-level guidance is what actually governs attorney conduct.

What This Means for Multi-State Operations

If your firm or legal department operates across jurisdictions, the compliance challenge is real. A workflow that is perfectly acceptable in Texas might violate Florida's disclosure requirements. An AI tool that passes muster under California's competence standard might not satisfy Colorado's new definitional framework. And court-specific standing orders add yet another layer, because your obligations can change depending on which judge you are in front of.

The practical upshot is that multi-state legal operations need to default to the most restrictive applicable standard, at least until the landscape settles. That means written client disclosures, documented AI tool vetting, independent verification of all AI-generated work product, and clear internal policies on billing.

How FirmAdapt Addresses This

FirmAdapt's architecture is built around jurisdiction-aware compliance controls, which means the platform can apply different disclosure, logging, and verification requirements depending on the state bar rules that govern a particular matter. When Florida requires written AI disclosure and Texas does not, FirmAdapt handles that distinction at the workflow level rather than leaving it to individual attorneys to remember.

The platform also maintains a continuously updated regulatory mapping of state bar guidance, court standing orders, and ABA opinions, so compliance teams can audit their AI usage against current requirements without manually tracking 50 different jurisdictions. For legal departments and firms operating across state lines, this kind of infrastructure is becoming a baseline operational need rather than a nice-to-have.