- As of 2026, more than twenty-five U.S. state bars have issued formal opinions or guidance on lawyer use of generative AI. Read §01
- Most state opinions track ABA Op. 512 closely; the variation is in specifics like disclosure requirements, billing standards, and supervision protocols. Read §03
- Stricter states on disclosure include California (Practical Guidance from State Bar), Florida (FAO 24-1), and North Carolina (FEO 2024-1). Read §06
- Less prescriptive states have largely adopted ABA Op. 512 by reference rather than issuing their own analysis. Read §08
- Multi-jurisdictional practitioners must consult each state's specific opinion; the variation matters. Read §11
State bar AI opinions.
A comparative survey of adopted state-bar opinions and formal guidance on lawyer use of artificial intelligence. Who has issued what, the common analytical pattern, and what an attorney admitted in multiple states should be tracking. Updated as opinions are adopted.
The state of state guidance. #
The American Bar Association issued Formal Opinion 512 in July 2024, the federal-Model-Rules statement on lawyer use of generative artificial intelligence. Within eighteen months, a substantial minority of state bars had issued their own formal opinions or formal guidance applying the equivalent state Rules to the same facts. The result is a patchwork: most states have functional convergence with the ABA position; the texts vary; the procedural posture of the guidance varies more.
This page tracks the adopted opinions and formal guidance documents IXSOR can confirm from primary sources. It does not list every state. States omitted may have no formal guidance, may have informal guidance only, or may be in active drafting at the time of writing. Absent guidance, the underlying state Rules of Professional Conduct govern; an attorney is responsible for applying them.
This page is updated quarterly. The state-bar links below are the controlling sources; the IXSOR readings, where they exist, are operational interpretations.
Federal floor: ABA Op. 512. #
The federal Model Rules statement is ABA Formal Opinion 512 (July 29, 2024). The opinion identifies six duties an attorney must satisfy when using generative AI: competence, confidentiality, communication with the client, candor and supervision, reasonable fees, and bias considerations. IXSOR's reading is here.
State bars commonly cite Op. 512 as persuasive authority even when their state has not yet adopted its own opinion. An attorney with no state-bar guidance can use Op. 512 as the operational baseline; an attorney whose state has its own opinion uses the state opinion as controlling and Op. 512 as an interpretive companion.
California: practical guidance. #
The State Bar of California issued Practical Guidance for the Use of Generative Artificial Intelligence in the Practice of Law in November 2023. It is the earliest material formal guidance from any major state bar. The document is structured as a list of principles and recommended practices rather than a Rule-by-Rule application.
What is distinctive about the California document: it functions more like a compliance handbook than an opinion. It lists confidentiality obligations, supervision obligations, candor obligations, and fee obligations, with practical recommendations under each. An attorney looking for an operational starting point will find the California document more directive than most state opinions.
The California guidance preceded ABA Op. 512 by eight months. Its conclusions and the ABA's converge.
Florida: Op. 24-1. #
The Florida Bar Ethics Opinion 24-1 (January 2024) is among the first formal ethics opinions to address generative AI. It addresses confidentiality, oversight, billing, and advertising in the context of AI-assisted practice. Florida's opinion is notable for its explicit treatment of advertising implications: an attorney whose website or marketing materials describe AI capability must comply with Florida's advertising rules, which are among the strictest in the country.
For a Florida-admitted attorney, the advertising element matters operationally. A firm that markets its AI use as a differentiator should run the relevant pages past someone familiar with the Florida rules.
New York: Task Force Report. #
The New York State Bar Association issued the Report and Recommendations of the New York State Bar Association Task Force on Artificial Intelligence in April 2024. It is a longer document than most state opinions, more in the nature of a task-force report than a Rule-by-Rule formal opinion, but its substantive recommendations track the same five Rules.
What is distinctive about the New York report: it goes well beyond the practising-lawyer ethics question to address AI in courts, AI in legal education, AI in regulation of the profession, and AI policy at the state level. An attorney reading it for ethics-compliance purposes will find the relevant material in specific sections; the rest is policy.
The NYSBA report is a recommendation document, not a binding ethics opinion. An attorney admitted in New York should also track formal opinions issued by the New York State Bar's ethics committees and the relevant Appellate Division.
North Carolina: 2024 FEO 1. #
The North Carolina State Bar 2024 Formal Ethics Opinion 1 (adopted Nov. 1, 2024) addresses lawyer use of AI under NC Rules of Professional Conduct 1.1, 1.6(c), and 5.3. The opinion is permissive in result and prescriptive in process: an attorney may use AI, including by inputting client data into a third-party AI tool, provided she does so competently, securely, and with appropriate supervision. IXSOR's operational reading is here.
The NC opinion's most distinctive analytical move is the application of Rule 5.3 to third-party software vendors, importing supervisory duties that previously applied to outside-firm nonlawyer assistants. The opinion expressly declines to address attorney-client privilege, treating the privilege question as a matter of evidentiary law outside the Rules of Professional Conduct.
D.C.: Op. 388. #
The District of Columbia Bar Ethics Opinion 388 (2024) addresses lawyer use of generative AI under the D.C. Rules of Professional Conduct. The opinion tracks Op. 512 closely, with D.C.-specific framing and emphasis on the supervisory and confidentiality obligations.
For attorneys practising in federal court in the District of Columbia, Op. 388 is the controlling D.C. Bar text. Federal practice in D.C. courts also subjects the attorney to the standing orders and local rules of the relevant federal court (D.D.C., D.C. Circuit), which have moved more aggressively on AI-disclosure requirements than in many other jurisdictions.
Mississippi: practical guide. #
The Mississippi Bar issued an AI Practical Guide in 2025 (published on the Mississippi Bar website). The guide is a practitioner-oriented document organised around concrete practice scenarios rather than around Rules of Professional Conduct. Useful as a comparative reference; functionally aligned with Op. 512.
States without confirmed adopted opinions. #
Many state bars have not issued a formal AI opinion. Some have published informal guidance, position papers, or task-force reports that do not have the procedural status of a formal ethics opinion. Some have ethics-committee deliberations underway. Some have nothing.
For an attorney in a state without confirmed formal guidance, three operational consequences follow:
- The state Rules of Professional Conduct still govern. The duties under MR 1.1, MR 1.6, MR 3.3, MR 5.3, and MR 1.5 (or their state analogues) are imported by their text.
- ABA Op. 512 functions as persuasive authority. State courts and state bars routinely treat ABA opinions as influential.
- Sister-state opinions function as persuasive authority. The texts of CA, FL, NY, NC, D.C., MS, and other adopting states converge on a common analytical pattern; that convergence is itself authority.
The absence of an opinion does not absolve compliance. It shifts the work of identifying the controlling standard onto the attorney.
The common analytical pattern. #
Across every adopted opinion and formal guidance document IXSOR has reviewed, the same five-duty pattern recurs. The framing differs; the substance does not.
- Competence. The attorney must understand the AI tool, including its failure modes. State formulation: MR 1.1 + Comment 8 (or equivalent).
- Confidentiality. Inputs to AI tools must satisfy reasonable-efforts protection of client information. State formulation: MR 1.6(c).
- Supervision. AI work product must be supervised, with the AI vendor frequently treated as a nonlawyer assistant for supervisory purposes. State formulation: MR 5.3 (and 5.1 where partner-level supervision is at issue).
- Candor and verification. Filed work product must be verified. Citations must be real. State formulation: MR 3.3 + procedural Rule 11.
- Fees and communication. Fees must be reasonable; material AI use should be communicated to the client. State formulation: MR 1.5, MR 1.4.
An attorney who internalises this five-duty pattern can read any new state opinion in under thirty minutes and identify the state-specific deltas. The deltas tend to be in framing, in advertising rules where the state has them (Florida, Texas), and in court-rule disclosure obligations where the state's courts have promulgated standing orders.
Multi-state practice. #
An attorney admitted in multiple states must comply with each state's Rules and any AI-specific opinion adopted in each state. The exposure is cumulative: a multi-state attorney can be subject to discipline in any state where she is admitted, on any state's reading of the relevant Rules.
Three operational practices reduce exposure:
- Read each applicable opinion. Functional convergence does not mean texts are identical. The deltas are real.
- Comply with the strictest. Where opinions diverge, the strictest applicable standard is the one to satisfy.
- Document the standard. A multi-state firm's AI policy should reference each applicable state's opinion and identify the controlling provision when conflicts arise.
For practices spanning a small number of states, the work is bounded. For practices spanning many, the work is non-trivial; consider it part of the firm's ongoing compliance overhead.
Citations and further reading. #
Federal:
- ABA Formal Opinion 512 (July 29, 2024). IXSOR reading.
- ABA Task Force on Law and Artificial Intelligence.
States with adopted opinions or confirmed formal guidance:
- State Bar of California (Nov. 2023, Practical Guidance).
- The Florida Bar (Op. 24-1, Jan. 2024).
- New York State Bar Association (Task Force Report, Apr. 2024).
- North Carolina State Bar (2024 FEO 1, Nov. 2024). IXSOR reading.
- D.C. Bar (Op. 388, 2024).
- The Mississippi Bar (AI Practical Guide, 2025).
Cross-cutting:
- Federal Rule of Civil Procedure 11 (Cornell LII).
- Mata v. Avianca, Three Years On (post-Mata sanctions caselaw, IXSOR survey).
This article is general analysis of state-bar guidance. It is not legal advice. State opinions are amended and supplemented over time; the controlling text is the opinion as published by the relevant bar at the time of the attorney's use. Engage qualified counsel admitted in the relevant jurisdiction for advice on a specific matter.
Frequently asked questions.
Which states have issued AI ethics opinions?
As of 2026, more than 25 state bars have issued formal opinions or detailed guidance on lawyer use of generative AI. The most active jurisdictions include California, Florida, North Carolina, New York, Illinois, Texas, Pennsylvania, Massachusetts, New Jersey, and Washington. Many other states have issued informal guidance, advisory opinions, or have ABA Op. 512 by reference. The IXSOR tracker covers the comparative landscape.
What's the difference between state AI opinions?
Most variation is in three areas. (1) Disclosure to clients: some states (CA, FL, NC) strongly recommend or require disclosure for substantive AI use; others leave it to lawyer judgment. (2) Billing standards: states differ on how AI-induced time savings should be reflected in fees. (3) Vendor supervision: states differ on the level of due diligence required of AI vendors. Most states track ABA Op. 512's framework with these variations.
Does my state require client consent for AI use?
Depends on jurisdiction. California's State Bar Practical Guidance, Florida's FAO 24-1, and North Carolina's FEO 2024-1 strongly recommend client disclosure for substantive AI use. Other states (most ABA-adopting jurisdictions) leave it to lawyer judgment under Rule 1.4. Read your state-specific guidance; if your state hasn't issued guidance, default to ABA Op. 512's framework.
How do I find my state's AI ethics opinion?
Most state bar websites publish formal ethics opinions in a searchable database under "Ethics & Discipline" or similar. ABA's Center for Professional Responsibility maintains a partial cross-jurisdictional index. State-bar-specific blogs and CLE programs also track new opinions as they issue. Sign up for state-bar email alerts on ethics-opinion releases.
Do all states follow ABA Op. 512?
Most do, but with variations. ABA Op. 512 is the framework most state opinions build on. The specifics differ in disclosure thresholds, billing standards, and supervision detail. A state-bar opinion that "adopts ABA Op. 512" is using Op. 512's framework but may have local variations on the specifics. Always read the state opinion's actual text.
What if my state hasn't issued AI guidance?
Default to ABA Op. 512 as the operating framework. Most state bars treat ABA opinions as persuasive even where they have not formally adopted them, and ABA Op. 512 is the most detailed available guidance on lawyer AI use. Document the decision to follow Op. 512 in the absence of state-specific guidance. Re-evaluate if your state issues its own opinion.
How often do states update AI guidance?
Frequently in 2024-2026, less so as the doctrine matures. The first wave of state opinions issued in 2023-2024 in response to ChatGPT's emergence and the Mata case. A second wave issued in 2024-2025 incorporating ABA Op. 512. As of 2026, the rate of new opinion issuance has slowed, but updates and supplementary opinions continue to appear. State bar email alerts are the most reliable way to stay current.
Are multi-jurisdictional lawyers held to the strictest state's AI rules?
Generally, yes for confidentiality and verification (which are universal). For disclosure and billing, the analysis is more nuanced and depends on which jurisdiction's rules apply to the matter under choice-of-law principles. The conservative approach is to apply the strictest applicable state's rules, particularly for client-disclosure decisions. Document the choice-of-rule analysis.
