Questions
Does the EU AI Act Apply to Law Firms? A Direct Answer
Yes. An EU law firm whose staff use ChatGPT, Claude or Copilot is a deployer under the EU AI Act. Article 4 literacy applies now, Article 50 since August 2026.
On this page
Yes. An EU law firm whose staff use ChatGPT, Claude, Copilot or Harvey is a deployer under the AI Act. As of September 2026 that means an AI-literacy duty under Article 4 (since 2 February 2025) and transparency duties under Article 50 (since 2 August 2026). High-risk obligations catch only specific uses and are deferred to 2 December 2027. Provider duties arise only if you build and market a system.
Deployer, provider or neither: the role decides everything
The Act does not regulate lawyers; it regulates roles. A provider develops an AI system or general-purpose model and places it on the market under its own name. A deployer uses an AI system under its own authority for professional purposes. OpenAI, Anthropic, Microsoft, Harvey and Legora are providers. Your firm, the moment a trainee opens Claude to summarise a lease, is a deployer.
Three things follow. The heaviest obligations, from model documentation to the AI Office's power since 2 August 2026 to evaluate models and fine up to €15m or 3% of global turnover, land on your vendors (Help Net Security, 4 Aug 2026). A firm that sells its own branded tool to clients moves towards the provider side. And scope follows use, not the brass plate: a London or New York firm with an EU office, or whose AI output is used in the Union, is inside the Act for that activity.
What applies to a law firm today, as of September 2026
| Provision | Applies since | Who it hits in a firm | After the Digital Omnibus |
|---|---|---|---|
| Art. 5 prohibitions | 2 Feb 2025 | Rare in legal practice | In force; two new prohibitions from 2 Dec 2026 |
| Art. 4 AI literacy | 2 Feb 2025; national supervision since 2 Aug 2026 | Every firm whose staff use AI | Obligation of effort; not deferred |
| GPAI model duties | 2 Aug 2025 | Vendors only | AI Office enforcement since 2 Aug 2026 |
| Art. 50 transparency | 2 Aug 2026 | Client-facing chatbots; published synthetic content | Not deferred; watermarking grace to 2 Dec 2026 |
| Annex III high-risk (justice, employment) | 2 Dec 2027 (was 2 Aug 2026) | HR screening; tools built for courts | Deferred by Regulation (EU) 2026/1744 |
| Annex I product-embedded high-risk | 2 Aug 2028 | Not legal practice | Deferred |
The Digital Omnibus on AI, Regulation (EU) 2026/1744, was published in the Official Journal on 24 July 2026 and has been in force since 27 July 2026 (Cloud Security Alliance). Nils Rauer of Pinsent Masons called the deal "a pragmatic step towards greater legal certainty" (Out-Law); critics call a 16-month delay plus a softer Article 4 deregulation by another name. The practical reading for a firm is the CSA's: deferred, not cancelled (full list of changes).
Article 4: the duty that already applies to you
The Commission's AI-literacy FAQ, updated 27 July 2026, settles scope: Article 4 applies to any organisation whose staff use tools like ChatGPT (European Commission). No certificates, no tests, but keep an internal training record; the Commission's repository lists 40+ practices to borrow from.
The Omnibus reworded the duty from ensuring "a sufficient level" of AI literacy to "take measures to support the development of AI literacy" (artificialintelligenceact.eu). That is an obligation of effort with no dedicated EU fine tier; national market-surveillance authorities have supervised it since 2 August 2026 and penalties are national and "proportionate". The €7.5m figure some compliance vendors quote is the tier for supplying incorrect information to authorities.
Do not read that as optional; the bars and courts bite harder. The Divisional Court in Ayinde v Haringey [2025] EWHC 1383 (Admin) addressed heads of chambers and managing partners directly on their responsibility for lawyers' use of AI (Burges Salmon), a court-made cousin of Article 4 in a jurisdiction the Act does not reach. In Austria the KI-Servicestelle at RTR (§ 20c KOG) runs the national KI-Kompetenz hub (RTR), and ÖRAK's four practice rules fit on a business card: EU servers, disclose AI use, the lawyer keeps final responsibility, no client data for training. Wolf Theiss's Chief Digital & Technology Officer Leonhard Wassiq told Kurier in 2026 that AI is "a question of organisation, not technology"; that is Article 4 in a sentence. See what Article 4 expects and what to teach.
Article 50: if a chatbot talks to your clients
Since 2 August 2026 a person interacting with an AI system must be told so, deepfakes must be labelled and synthetic content must carry machine-readable marks; providers of existing generative systems have until 2 December 2026 to finish the watermarking (Gibson Dunn). Anthropic reportedly added a statistical watermark to Fable 5.1 outputs for this purpose.
For a firm the live case is the intake chatbot on the website. The design duty sits with the provider, but the bot carries your name and faces your prospective client, so put the disclosure in its first message whether or not you are the provider on paper. Memos and letters to a client are not caught; your duty to tell clients about AI use comes from the professional rules, not Article 50.
High-risk: parked until December 2027, but not for every use
Annex III lists AI intended to assist a judicial authority in researching and interpreting facts and law. Whether a lawyer-side research tool is caught is unsettled, and the Omnibus has parked the question until 2 December 2027. Two uses inside a firm are clearer. Employment is an Annex III area, so an AI tool that screens CVs for the trainee intake makes the firm a deployer of a high-risk system from that date, with duties of human oversight, logging and informing affected workers (HR rules). And a firm that puts its name on a high-risk system, or substantially modifies one, takes on provider duties under Article 25. Fines for most breaches run to €15m or 3% of worldwide turnover; €35m or 7% for prohibited practices.
What to do this quarter
- Map the systems. List every tool with a firm licence or known unofficial use: Copilot, Claude, ChatGPT, Harvey, meeting recorders, the website chatbot. Note whether each touches HR, courts or the public. Thomson Reuters found 34% of professionals use unsanctioned "shadow AI" (June 2026); the list is longer than the licence register.
- Write the Article 4 record. One page per role (partners, associates, support staff, IT): what training, when, who attended, where the tool fails. Date it. The Commission asks for nothing more formal.
- Fix the chatbot. Add the AI disclosure now. The Article 50 grace period covers providers' watermarking, not your bot's first message.
- Flag the HR use. If any recruitment step uses AI ranking, diarise 2 December 2027 and design the human oversight in 2027, not 2028.
- Fold it into the AI policy. GDPR runs alongside the Act, and EDPB Opinion 28/2024 asks deployers to assess upstream unlawful training; the policy template and the GDPR page give you the wording.
Module 6 of the AI Lab for Lawyers course, on risks and regulation, walks through this mapping and the Article 4 record. The complete guide to the EU AI Act for law firms goes article by article.
Frequently asked questions
Is a law firm a deployer or a provider under the EU AI Act?
Almost always a deployer. A firm that uses Claude, ChatGPT, Copilot, Harvey or Legora under its own authority for professional purposes is a deployer; OpenAI, Anthropic, Microsoft and the legal-tech vendors are the providers. A firm becomes a provider only where it develops a system and places it on the market under its own name, or puts its name on a high-risk system or substantially modifies one, which is a case-by-case analysis.
Does the EU AI Act apply to a UK or US law firm?
Not to a firm with no EU establishment whose AI output stays outside the Union. The Act reaches providers and deployers in third countries where the output produced by the system is used in the EU, so a London or New York firm with a Frankfurt or Vienna office, or serving EU matters with AI-generated work product, is inside the Act for that activity. UK firms still face the SRA warning notice of 17 August 2026 regardless.
What does Article 4 AI literacy actually require a law firm to do?
Take measures to support the development of AI literacy among the people who operate or use AI systems on the firm's behalf. The Commission FAQ updated on 27 July 2026 says this covers any organisation whose staff use tools like ChatGPT, requires no certificates or tests, and expects an internal training record. National market-surveillance authorities have supervised the duty since 2 August 2026; there is no dedicated EU fine tier for it.
Are legal research and drafting tools high-risk under the EU AI Act?
Unsettled, and parked until 2 December 2027. Annex III lists AI intended to assist a judicial authority in researching and interpreting facts and law, and commentators disagree on whether lawyer-side research tools fall within it. The Digital Omnibus deferred all stand-alone Annex III obligations from 2 August 2026 to 2 December 2027, so no research tool carries high-risk duties as of September 2026. AI used to screen job applicants is the clearer high-risk case for a firm.
Put this into practice
AI Lab for Lawyers: a live, hands-on course for practising lawyers
4 live sessions (8 hours), recordings included, taught by Dr. Niklas Schmidt of Wolf Theiss. Rated 4.7/5 from 381 ratings; alumni from Norton Rose Fulbright, Paul Hastings, Pillsbury and Taylor Wessing.
Related reading
Regulation & Ethics
The EU AI Act for Lawyers and Law Firms: What Applies in 2026
The EU AI Act applies to law firms as deployers. Article 4 literacy and Article 50 transparency are live; high-risk duties are deferred to December 2027.
9 min read
Regulation & Ethics
EU AI Act Article 4: The AI Literacy Duty for Law Firms, Explained
Article 4 of the AI Act requires every firm whose staff use AI to support their AI literacy. In force since 2 February 2025, supervised nationally since August 2026.
7 min read
Regulation & Ethics
The Digital Omnibus on AI: What Changed in the AI Act and What Did Not
The Digital Omnibus AI Act amendment, Regulation (EU) 2026/1744, deferred high-risk duties to December 2027 but left transparency, literacy and the prohibitions in force.
6 min read
Glossary
Deployer (EU AI Act) — Definition for Lawyers
A deployer under the EU AI Act is anyone using an AI system under its own authority for professional purposes. Law firms are deployers; here is what follows, and when.
4 min read
Regulation & Ethics
AI Rules for Austrian and German Lawyers: ÖRAK, BRAK and CCBE Guidance
ÖRAK's four AI rules and § 40 Abs 3 RL-BA bind Austrian lawyers; BRAK's § 43a BRAO line binds German ones; CCBE guides cover both. With a bilingual checklist.
11 min read
Regulation & Ethics
The SRA's 2026 Warning Notice on AI Misuse: What UK Solicitors Must Know
The SRA AI warning notice of 17 August 2026 followed 42 misuse reports in a year. Solicitors must understand a tool well enough to review its output critically.
8 min read