Technology CLE Requirements by State: AI Credit, What Counts and What Is Mandatory

Technology CLE requirements rarely make the news; sanctions orders do. The fastest route to a mandatory AI CLE in 2026 is to file a brief you did not read. On 9 September 2026 a federal judge in Utah ordered the lawyers in Beus Gilbert v. BYU to complete two AI-ethics CLE courses alongside a $3,000 fine, after a filing built with ClearBrief, Claude, ChatGPT and Gemini contained one fabricated and three misrepresented citations. In June, Withers v. City of Aberdeen (N.D. Miss.) added an AI-ethics CLE to fines and disqualification for lawyers on both sides of the same case.

Outside the courtroom the position is quieter than the headlines suggest. As of September 2026 only three states mandate technology CLE hours, and none mandates AI-ethics CLE, although 67% of Texas lawyers want one. What every US lawyer does have is a duty of technology competence, adopted in 40 states, which ABA Formal Opinion 512 has applied to generative AI. Here are the requirements as they stand, what counts, and how to turn the AI training you should be doing anyway into credit.

The state table

The state bar AI ethics opinions map covers the opinions; this page is about the hours.

Comment 8: technology competence in 40 states

Comment 8 to Model Rule 1.1 was added in 2012 and has been adopted by 40 states plus DC and Puerto Rico. It imposes no hours. What it does is make “I did not understand the tool” an admission rather than a defence.

ABA Formal Opinion 512 (29 July 2024) is the AI gloss: lawyers “need not become GAI experts” but “must have a reasonable understanding of the capabilities and limitations” of the tools they use, and supervising lawyers must ensure “subordinate lawyers and nonlawyers are trained”. Texas Opinion 705 adds the word “current”, which does real work with models that change quarterly. The Opinion 512 explainer covers the rest.

Mandatory technology CLE: Florida, North Carolina, California

Florida went first: 3 hours per three-year cycle, approved 29 September 2016, effective 1 January 2017. Bill Hamilton of the University of Florida’s E-Discovery Project told Bloomberg Law that “Three hours isn’t enough to turn someone into a technologist”, and also that “The hours aren’t a burden, they are a form of liberation.” The total fell back from 33 to 30 hours in January 2024; the 3 technology hours stayed.

North Carolina followed with 1 hour a year from 2019. Its definition is the most useful in the country because it plainly covers a hands-on AI session: a programme “devoted to education on information technology (IT) or cybersecurity, including education on an information technology product, device, platform, application, or other tool, process, or methodology”.

California requires at least one hour on technology in the practice of law under Rule 2.72©(2)(a)(iv).

Proposed AI-ethics CLE: Texas, and the courts that order it

No state has yet made AI-ethics CLE mandatory. The pressure comes from two directions.

From lawyers: in the State Bar of Texas 2026 survey of 1,553 lawyers, 67% supported required CLE on the ethical use of AI, down from 71% in 2024 but still the most-supported regulatory option, while only 9% were “extremely confident” evaluating AI-generated documents. The comments are specific about format: “watch-and-learn, not lecture and static powerpoint slides” and “less CLE regarding the ethics of AI use and more about effective, safe, and practical uses”.

From the bench: Dehghani v. Castro (D.N.M., April 2025) imposed a fine, mandatory CLE and self-reporting on an immigration lawyer whose freelancer had used generative AI; Withers and Beus Gilbert followed in 2026. The Ninth Circuit in Lnu v. Blanche (3 June 2026) went past CLE to conduct: $2,500 each, six-month suspensions from Ninth Circuit practice, and two years of under-penalty-of-perjury disclosures naming any AI tools used and certifying personal citation review, because “A competent and diligent attorney must also read and reason.”

What counts as technology CLE

The definitions are broader than most lawyers assume: North Carolina’s covers any “tool, process, or methodology”, California’s “technology in the practice of law”. A hands-on session on prompting, confidentiality settings and citation verification fits both. The accredited examples, as of September 2026:

The free vendor certifications from Clio and Harvey Academy state no CLE at all; “free certification” and “CLE credit” are different products, and the comparison of AI courses for lawyers separates them. The credit column is dominated by California because providers apply there first; ask which other states a provider has sought.

Getting AI training accredited, and ethics credit for AI courses

If you run training, the accredited courses share three visible features: a lawyer instructor, a published syllabus with stated hours, and a technology-versus-ethics split that maps to the state’s categories; Berkeley’s 4.5 plus 0.5 is the model. Each state accredits separately.

If you attend training, keep the paperwork even when the course is not accredited. Some states accept self-reported credit; all of them, and your insurer, will accept a training record as evidence of competence. Underwriters have started asking “Do you use AI? Do you police it? Do you have protocols in place?” (Aon’s Stan Sterna, April 2026), and training records sit alongside a pre-filing verification log on the list of renewal artefacts to keep, as the malpractice insurance guide explains.

Here is the certificate of completion and the agenda for a course I attended: <certificate>...</certificate> <agenda>...</agenda>.
Produce a one-page training record with these fields: dates; provider and instructor; format (live / self-paced / hands-on); hours by category (technology / ethics / general) as stated by the provider, or "not accredited"; tools covered; the three skills practised; the verification routine taught, in one sentence; jurisdictions in which credit is claimed and the rule relied on, tagged [VERIFY]; and a "next refresh due" date twelve months out.
Do not infer credit the documents do not state. Plain text I can file.

On ethics credit: the sanction orders call the required course “AI-ethics CLE”, and most accredited AI courses include a small ethics component. The Texas comments are a warning to providers: lawyers have had the lecture. The ethics hour that gets watched is the one where a fabricated citation is produced live on screen and then checked in a real database.

I am licensed in [state(s)]; my cycle ends [date]; my technology requirement is [3 hours per 3 years / 1 hour per year / at least 1 hour / none, but Comment 8 applies]. I use [ChatGPT Business / Copilot / a legal research platform] for [tasks].
Recommend how to spend the hours so they also close my biggest competence gaps: one session on how models fabricate and how to verify citations, one on confidentiality settings and tiers, one hands-on session on my own document types. For each, say what evidence to keep. Do not name courses; give me criteria and the questions to ask a provider about accreditation in my state.

For a filing in [court, judge], using only the court's website <url> and the attached order <order>, tell me whether any generative-AI standing order, certification, disclosure or training requirement applies, quoting the operative words. If this court has sanctioned lawyers for AI-fabricated citations, list the conditions it imposed (CLE, disclosure, self-reporting). If you find nothing in the sources provided, say "NO ORDER FOUND ON THE SOURCES PROVIDED" rather than guessing.

Run that before you file anywhere new; the court standing orders guide tracks the orders.

Beyond the US: SRA competence and Article 4 records

England and Wales has no CLE hours; solicitors have a continuing competence duty, and the SRA’s warning notice of 17 August 2026 attaches AI to it, citing Code paragraph 3.6, keeping “professional knowledge and skills… up to date”. Its line on outcomes: “Reliance on an output of AI would not be a suitable defence.” The SRA warning notice guide sets out what the regulator now expects.

European Union. Article 4 of the AI Act has applied to every deployer since 2 February 2025, and the Commission’s Q&A is the nearest thing to a European CLE rule: no certificate, no AI officer, no mandated level, but “Organisations can keep an internal record of trainings” and national sanctions are “more likely if there is proof of an incident due to lack of appropriate training”. Austria’s ÖRAK goes further: AI use is permitted only where staff have “KI-Kompetenz”. The Article 4 guide has the record template.

The through-line, from Florida’s three hours to the Commission’s internal record, is the same: nobody counts your hours as closely as they will count your citations. AI Lab for Lawyers is eight live hours of the hands-on work North Carolina’s definition describes, with a certificate of completion for your file; it lists no state accreditation, so treat it as competence first and self-reported credit where your rules allow.

Where to go next: the AI training guide for what the hours should contain, the course comparison for who carries credit, and the learning paths hub for the rest.