The Legal AI Bar Compliance Gap Report 2026: How State Bar AI Guidance Has Evolved in the First Half of 2026 — and How Far Actual Firm Policy Lags Behind

Research briefing — a review of public sources. The Legal Stack did not conduct a survey, benchmark or interviews for this piece.

Executive summary and correction

An earlier version of this briefing reported that "thirty-one state bars" had issued AI guidance and that seventeen were "silent." It also cited a set of formal opinions: California Formal Opinion 2025-1, NYC Bar Formal Opinion 2025-4, Texas Opinion 701 as an AI opinion, and Virginia LEO 1912. We could not verify those counts or opinion numbers, and several do not exist as described. They have been removed. Texas Opinion 701, for example, concerns subscription fee arrangements; Texas's generative AI opinion is Opinion 705. We hold no dataset of firm AI policies, so the earlier "compliance floor" scores were not measurements either.

The verifiable picture is still significant. Since late 2023, a growing set of bars and courts has applied existing professional-conduct rules to generative AI. In 2026 the trend moved from guidance toward binding rule text. California is the most important development for multi-jurisdictional firms, and Florida now has a statewide certification rule for court filings. On the policy side, the best public survey data suggest many organizations have no AI policy at all. That gap is the more pressing problem.

The verified landscape

The opinions below are listed in public compilations, including Netforlawyers' running list of ethics opinions on generative AI and LawSites' February 2025 compendium. They are summarized from those secondary sources and from bar materials we reviewed.

Pre-2025 foundation

California State Bar, Practical Guidance for the Use of Generative AI (November 2023). Non-binding guidance, now the basis for proposed rule amendments (below).
Florida Bar Ethics Opinion 24-1 (January 2024). LawSites' compendium summarizes it as follows. Informed consent is "advisable" before using a third-party AI tool that would disclose confidential information. AI output must be reviewed, since lawyers otherwise face "the same perils as relying on an overconfident nonlawyer assistant." Lawyers may not bill time saved by AI as if they had done the work.
New Jersey Supreme Court, Preliminary Guidelines (January 2024). These guidelines say there is "no per se requirement to inform a client" about AI use. Disclosure is needed where nondisclosure would prevent informed client decisions.
Kentucky Bar Association Ethics Opinion E-457 (March 2024). There is no general duty to disclose "rote" AI-assisted research. Fees may need to fall where AI significantly cuts time, and AI tool costs may be charged only if the client agrees in writing.
D.C. Bar Ethics Opinion 388 (April 2024) addresses competence, confidentiality, supervision and fees.
Pennsylvania Bar Association and Philadelphia Bar Association Joint Formal Opinion 2024-200 (2024). Lawyers "remain fully accountable for the results" of AI-assisted work and should not bill AI-assisted work as if it were done manually.
ABA Formal Opinion 512 (29 July 2024). The national baseline, covering competence, confidentiality, communication, candor, supervision and reasonable fees. It is advisory and binds no state.
NYC Bar Formal Opinion 2024-5 (7 August 2024). Informed client consent is required before confidential information is entered into a tool that stores or shares inputs outside the firm. Firms should adopt AI policies and training.
North Carolina 2024 Formal Ethics Opinion 1 (November 2024). Stops short of requiring client consent. It cautions about self-learning tools and requires lawyers to exercise independent judgment over AI output.

2025 and after

Texas Professional Ethics Committee Opinion 705 (February 2025). The opinion was finalized after a November 2024 draft (Proposed Opinion 2024-6). The State Bar of Texas's AI toolkit summarizes it this way: lawyers must maintain technological competence, independently verify AI output before use, protect confidential information, and bill only for actual time, with AI efficiencies reflected in invoices.
Illinois published The Illinois Attorney's Guide to Implementing AI in late 2025, according to Netforlawyers' list.

2026: from guidance to rules

California. The California Supreme Court wrote to the State Bar's ethics committee (COPRAC) on 22 August 2025, asking it to consider writing the 2023 guidance into the formal rules. LawSites reported on 5 May 2026 that COPRAC approved proposed amendments to six rules on 13 March 2026:

Rule 1.1: a new comment requiring a lawyer to "independently review, verify, and exercise professional judgment regarding any output" of technology used in a representation, with no carve-out for routine work.
Rule 1.4: a communication duty where technology use "presents a significant risk or materially affects the scope, cost, manner, or decision-making process" of the representation. This is not a blanket disclosure mandate.
Rule 1.6: treating exposure of confidential information to AI tools as potentially "revealing" it where there is a material risk of inconsistent access, retention or use.
Rule 3.3: a duty to "verify the accuracy and existence of cited authorities."
Rules 5.1 and 5.3: extending managerial and supervisory duties to AI use.

The comment period closed on 4 May 2026. The Supreme Court has final authority, and no effective date has been set.

Florida. According to Jones Day, the Florida Supreme Court amended Rule of General Practice and Judicial Administration 2.515 on 28 May 2026, effective 15 June 2026. Every signer of a Florida court filing, including self-represented litigants, now certifies that "all legal authorities exist and are accurately cited." A companion administrative order preempts the circuit-level AI disclosure orders that had built up.

We could not verify the earlier version's claims about pending rulemaking in Washington, Ohio, Massachusetts, Maryland, Arizona or Michigan, or the comment volumes it attributed to them. Readers should check those jurisdictions directly.

Where the guidance actually diverges

Most of the verified opinions share the same core: competence, confidentiality, verification of output, supervision and reasonable fees. The real differences are narrower than the earlier version suggested.

Client disclosure and consent. The trigger differs. NYC 2024-5 ties consent to tools that store or share inputs outside the firm. Florida 24-1 recommends informed consent when confidential information goes to a third-party tool. New Jersey and Kentucky reject a per se disclosure duty. California's proposal would use a risk-and-materiality test. A national firm that writes one disclosure trigger should test it against each.
Billing. Florida, Kentucky, Pennsylvania and Texas all address AI and fees. The common thread is no billing for time not spent, and charging AI tool costs only with client agreement (Florida and Kentucky specify writing).
Court-facing verification. This is now hardening into rules: Florida's certification and California's proposed Rule 3.3 comment.

No verified opinion we found addresses conflict-of-laws questions for AI disclosure on cross-border matters. Model Rule 8.5 remains the starting point; that is our analysis, not bar guidance.

How far firm policy lags (what the data does and does not show)

No public dataset scores firm AI policies against these opinions. The closest evidence is survey-based:

The Thomson Reuters Institute's 2025 Generative AI in Professional Services report (summarized 15 April 2025) found that 52% of the legal professionals surveyed said their organization had no GenAI usage policy, and 64% had received no GenAI-specific training.
The ACC/Everlaw survey of 657 in-house professionals (14 October 2025) found that 59% of in-house teams do not know whether their law firms use GenAI on their matters.

Even a good policy is not compliance. In Johnson v. Dunn (N.D. Ala., 23 July 2025), EDRM reports that Butler Snow already had AI policies requiring written approval and independent verification. Three of its lawyers were still disqualified and referred to the Alabama State Bar over fabricated citations. The court credited the policies in declining to sanction the firm itself.

Implications for ethics counsel and GCs (analysis)

Map your footprint, not a headcount of bars. List the jurisdictions where your lawyers are admitted and file. Then pull the specific opinions and rules, starting with California's proposals and Florida Rule 2.515.
Write verification into the workflow. Every verified source converges on independent verification of AI output. Build a documented pre-filing and pre-sending check, and assign it to the lawyer who signs.
Make the disclosure trigger jurisdiction-aware. Use the most demanding applicable standard on matters touching California, New York or Florida until the California rules are final.
Align billing policy with the no-billing-for-time-not-spent consensus.
GCs: Given the ACC finding, ask outside counsel for their AI policy and for confirmation of their verification practice when updating outside counsel guidelines.

This briefing is not legal advice.

Sources

Netforlawyers, "Ethics Opinions Covering Lawyers' Use of Generative AI (Artificial Intelligence)" (undated; accessed Oct 2026): https://www.netforlawyers.com/content/ethics-opinions-covering-lawyers-use-generative-ai-artificial-intelligence
LawSites (LawNext), "A compendium of legal ethics opinions on gen AI, as compiled by (you guessed it) gen AI" (Feb 2025): https://www.lawnext.com/2025/02/a-compendium-of-legal-ethics-opinions-on-gen-ai-as-compiled-by-you-guessed-it-gen-ai.html
Frankfurt Kurnit Klein & Selz, "ABA issues comprehensive formal ethics opinion on lawyers' use of generative AI" (27 Jan 2025): https://technologylaw.fkks.com/post/102jwnl/aba-issues-comprehensive-formal-ethics-opinion-on-lawyers-use-of-generative-ai
Ezel, "Texas Opinion 705: Generative AI in the practice of law" (listing dated 1 Feb 2025): https://ezel.ai/ethics-opinions/tx/705-generative-ai-in-the-practice-of-law
State Bar of Texas (Texas Bar Practice), "AI Guidance for Legal Professionals / Artificial Intelligence Toolkit" (accessed Oct 2026): https://www.texasbarpractice.com/artificial-intelligence-toolkit/
LawSites (LawNext), "California bar proposes rule requiring lawyers to verify every AI output and five other AI-focused ethics changes" (5 May 2026): https://www.lawnext.com/2026/05/california-bar-proposes-rule-requiring-lawyers-to-verify-every-ai-output-and-five-other-ai-focused-ethics-changes.html
State Bar of California, "Proposed Amendments to the Rules of Professional Conduct Related to Artificial Intelligence" (2026 public comment): https://www.calbar.ca.gov/public/public-meetings-comment/public-comment/public-comment-archives/2026-public-comment/proposed-amendments-rules-professional-conduct-related-artificial-intelligence
Jones Day, "Supreme Court of Florida creates statewide standard for artificial intelligence certification in court filings" (Jun 2026): https://www.jonesday.com/en/insights/2026/06/supreme-court-of-florida-creates-statewide-standard-for-artificial-intelligence-certification-in-court-filings
Thomson Reuters, "2025 Generative AI in Professional Services report: executive summary for legal professionals" (15 Apr 2025): https://legal.thomsonreuters.com/blog/genai-report-executive-summary-for-legal-professionals/
Association of Corporate Counsel, "ACC GenAI report: corporate law departments' AI use (with Everlaw)" (14 Oct 2025): https://www.acc.com/about/newsroom/news/acc-genai-report-corporate-law-departments-ai-use-everlaw
EDRM, "When AI Policies Fail: The AI Sanctions in Johnson v. Dunn and What They Mean for the Profession" (Jul 2025): https://edrm.net/2025/07/when-ai-policies-fail-the-ai-sanctions-in-johnson-v-dunn-and-what-they-mean-for-the-profession/