The Best AI Courses for Lawyers in 2026, Compared Honestly

Early in 2026 a Texas lawyer answered the State Bar’s AI survey with a request that should embarrass every CLE provider: “There needs to be more training available on how to use the programs in practice, similar to a lab you might have in biology class.” Search for the best AI courses for lawyers and the list that ranks recommends Harvard’s CS50, a UNESCO MOOC, DeepLearning.AI and Oxford’s general AI programme. None of them will show you how to review an NDA.

The demand is real. In the 8am 2026 Legal Industry Report, 69% of legal professionals personally use generative AI for work, yet 54% say their firm provides no training on responsible use and has no plans to. Law360 found under 15% of small-firm lawyers trained by their firm, against two-thirds in BigLaw. Here is the market as of September 2026, priced, and what each course is honestly good for.

Three tests before you pay for anything

Is it hands-on? GC AI’s May 2026 course guide says it in five words, “Self-paced is convenient. Live is effective”, and offers the test I would apply to any syllabus: “can I take a real contract from my current workload and apply what I learned before the end of this session?”

Is it tool-agnostic? Artificial Lawyer’s read on the January 2026 wave of vendor certifications is that they “help with client outreach, and reduce the burden on vendors to provide quite as much hands-on training”. Clio’s 2025 Legal Trends Report found nearly half of legal professionals use ChatGPT, Gemini, Claude or Perplexity and only 40% a legal-specific tool, so a one-platform course may teach the tool you do not have.

Was it rebuilt for 2026 models? OpenAI now says telling its reasoning models to “think step by step” is unnecessary; Anthropic says legacy “double-check your answer” lines “cause over-verification” in Opus 5. A 2024 recording teaches rituals that now hurt.

CLE credit matters less than people think: Bill Hamilton of the University of Florida’s E-Discovery Project said of Florida’s mandate, “Three hours isn’t enough to turn someone into a technologist.” Credit is now also a sanction: Beus Gilbert v. BYU (D. Utah, 9 September 2026) ordered two AI-ethics CLE courses. The technology CLE requirements guide covers the state rules.

Academic programmes: Berkeley, Stanford and Harvard

Berkeley Law Executive Education, “GenAI for the Legal Profession: Power User Edition”. Taught by Wei Chen, chief legal officer at Infoblox; self-paced; launched 16 February 2026; about eight hours in nine modules from foundations through research and drafting (“avoiding AI slop” is on the syllabus) to governance. $950, with discounts to $655-$855 for alumni, government, non-attorneys and groups; 6.5 California MCLE hours. The most complete self-paced syllabus on the market; nothing, however, makes you do the exercises.

Stanford Law, “AI Strategy for Legal Leaders”. About eight hours self-paced plus two live Zoom sessions, $900 introductory tuition, up to 8 California MCLE hours, Harvey access included, cohort 2 registration opening in autumn 2026. Aimed at general counsel: a strategy course for people who buy tools, not a lab for people who use them.

Harvard Law, “AI and the Law: Navigating the New Legal Landscape”. In person with the Berkman Klein Center, next dates TBD as of September 2026, no CLE credit listed; GC AI puts the price at $7,800. The most expensive option, and the one whose page says least about what you will do with your hands.

For calibration, Berkeley’s own JD course is more than 100 hours of hands-on work on Claude but “platform agnostic”; what law schools teach about AI shows how few practitioner courses match it.

Vendor academies: Harvey, Clio, and what Latham does in-house

Harvey Academy launched on 27 January 2026 as “a centralized learning hub designed to make legal AI literacy accessible to everyone”. Level 1 Foundations takes “about three hours, no Harvey license required”; its “AI for Legal Basics” module is 67 minutes and tool-agnostic. The free basics hour is worth an hour; the rest teaches Harvey. Clio’s Legal AI Fundamentals Certification is free, self-paced, about 2.5 hours of five short videos and a quiz, no CLE.

The firms that take this seriously build their own. Latham & Watkins ran a mandatory two-day AI Academy for its roughly 400 first-year associates in 2024 and again in 2025, and in August 2025 licensed Harvey for 3,600-plus lawyers with everyone to be trained by year end. Sidley’s London office added a mandatory “AI Knowledge Lab” seat to every training contract in September 2026, “a working seat, not a classroom”. If your firm does none of this, the AI training for lawyers guide explains how to build the equivalent.

MOOCs: Michigan’s specialisation on Coursera

“AI for Lawyers and Other Advocates” (Patrick Barry, University of Michigan) is four courses, about 16 hours, rated 4.6 from 86 reviews with roughly 5,100 enrolled; free to audit, no CLE. Two learner reviews tell you everything: “Practical course for professionals on delegating tasks to AI” and “Generalist introduction to using AI LLMs to help with writing. Doesn’t directly focus on legal tasks.” Both are right: take it for better emails, not for a contract-review workflow.

Free options worth your time, and the list that is not

GC AI’s free live classes (101 prompting, 201 advanced, 105 Word, playbooks, ethics) run 60 to 90 minutes with 1 to 1.25 California MCLE hours each, taught by former general counsel.

The list to skip is the one that ranks: legal-tech.blog’s “10 Best AI Training Courses for Lawyers” is CS50, a UNESCO MOOC, DeepLearning.AI and Oxford’s generic AI programme. Good for learning what a transformer is; no use for Tuesday’s NDA.

Free courses share one flaw: nobody checks the homework. Anisja Porschke in beck-aktuell: “Wer nur Inhalte konsumiert und sich nicht selbst mit praktischen Anwendungsfällen befasst, wird kein nachhaltiges KI-Verständnis aufbauen”, that is, consume without applying and nothing sticks. This prompt turns any free course into a lab.

Build me a six-week plan of 30 minutes a day to become competent with [ChatGPT Business / Claude Team / Copilot] for [transactional / litigation] work, alongside the [name of course] I am taking. Each day: one task using my own non-confidential or anonymised materials, the prompt pattern it teaches, a success criterion and one thing that could go wrong. Week 1: low-stakes personal tasks. Week 2: summarising and extraction. Week 3: drafting with verification of every fact. Week 4: review against a checklist. Week 5: a reusable Project. Week 6: one full workflow. Never put client data into a consumer tool. End with a self-assessment I can score.

Cohort courses on Maven: GC AI, the Atticus Project and AI Lab for Lawyers

Maven’s “AI for Lawyers” category listed eight cohort courses, one workshop and 17 free lightning lessons on 12 September 2026. Cecilia Ziniti (GC AI founder, three-time general counsel) teaches “AI Prompting for In-House Legal – 101”, 75 minutes live, $225, 4.6 from 445 reviews, and a 90-minute “201” at the same price: a strong first hour for an in-house lawyer, by design not a workflow course. The Atticus Project’s “AI Habits for Legal Professionals” (Wei Chen again) is a two-hour live workshop plus an email habit programme, 2 California MCLE hours.

AI Lab for Lawyers is mine, so weigh what follows accordingly. Four live two-hour sessions, recorded, more than 300 slides, homework, a community and a certificate, under the Maven Guarantee. The tools on screen are ChatGPT, Claude Cowork, Perplexity, NotebookLM and Harvey; the modules run from how LLMs work through daily legal work, prompt engineering and confidentiality to business development, platforms and agents. Nearly everything is browser-based and nobody codes. It is rated 4.8 from 320 ratings; participants have come from 60-plus firms including Norton Rose Fulbright and Paul Hastings, and reimbursement is often possible. The positioning is deliberate: “Many AI courses fall short: some drown you in the technical jargon of ‘neural networks’ and ‘transformers,’ while others, taught by lawyers, only skim the surface with a bit of basic ‘prompt engineering.’” The FAQ is equally candid: “Will you deal with AI risks and the regulation of AI? Not really.” If you need ethics credit, add a GC AI or Lawline session. The syllabus and every review are at maven.com/niklas-schmidt/ai-lab-for-lawyers.

Your firm may pay: the Maven FAQ anticipates “Can I get reimbursed by my firm?” and GC AI’s guide asks “How do I make the budget case for an AI course to my legal department head?”. This prompt drafts it.

Draft a one-page memo to [managing partner / general counsel] requesting approval for [course name, price, hours, format]. Structure: (1) the gap, using only these facts: 54% of firms provide no responsible-AI training (8am 2026 Legal Industry Report); ABA Formal Opinion 512 requires "a reasonable understanding of the capabilities and limitations" of the tools we use; (2) what I will bring back: a written playbook for [two named workflows], a verification checklist and a 45-minute session for [team]; (3) cost against [our hourly rate]; (4) how we will know in 90 days whether it worked. Under 400 words, no hype, no invented statistics.

German-language options and the Article 4 question

For Rechtsanwälte, Steuerberater and in-house Juristen the market is thinner and shaped by law. Anwaltverlag’s juracampus sells a self-study course, “Grundlagen der KI-Kompetenz (Schulung gemäß Art. 4 KI-VO)”, and digitalekanzlei.de a “KI-Kompetenzschulung im Sinne von Art. 4 KI-VO”, both built around the EU AI Act’s AI-literacy duty, in force since 2 February 2025 and softened by the Digital Omnibus in July 2026 to a duty to “take measures to support the development of” AI literacy. The Commission requires no certificate and no AI officer, but warns that “simply relying on the AI systems’ instructions for use or asking the staff to read them might be ineffective”. A live, documented, hands-on session is the kind of measure a firm can put on file. The BRAK, DAV and ÖRAK guide and the Article 4 explainer cover the rules; this prompt produces the record.

Draft an AI-literacy record for [law firm of N staff in Germany / Austria] covering the European Commission's minimum content: a general understanding of AI (what it is, how it works, which systems we use, dangers including hallucination); our role as deployer; the risk level of the systems we use; measures tailored to lawyers, paralegals, admin and IT. Include an inventory of AI tools in use, the training completed (course, date, hours, format, attendees), the next refresh date and where attendance evidence is stored. Plain German, one page.

The comparison table

Prices and hours as published on 12 and 13 September 2026; check the provider before booking.

Which course for which lawyer

• Solo or small-firm lawyer already using ChatGPT. GC AI’s free 101 for vocabulary, then a live cohort where you bring your own documents. Clio’s 2026 data says 86% of solos have not repriced after adopting AI; a course that never touches workflow and pricing will not change that.
• BigLaw associate. If your firm’s training was one webinar, the Paul Weiss lesson applies; the AI skills lawyers need lists what to learn next.
• General counsel. Stanford for strategy, GC AI for prompting, a workflow course for the rest.

Where to go next: the learning paths hub lists every guide on training and CLE; the best AI tools for lawyers says which tools a course should cover; the prompt library is where the exercises start. And if you want the lab the Texas respondent asked for, AI Lab for Lawyers is four two-hour sessions of exactly that.