Most of the EU AI Act is written in two voices. Some duties belong to the provider, the company that builds an AI system and puts it out under its own name. Others belong to the deployer, the company that uses an AI system in its own work. Read the Act without knowing which one you are, and every article looks like it applies to you.

The good news: for most companies the answer is clear once you look at one system at a time. The bad news: very few companies have the list of systems to look at.

The two definitions

The Act’s own words, shortened:

  • Provider (Article 3(3)): develops an AI system or has it developed, and places it on the market or puts it into service under its own name or trademark, whether for payment or free of charge.
  • Deployer (Article 3(4)): uses an AI system under its authority, except for purely personal, non-professional use.

The role belongs to the system, not to the company. The same company is often a deployer for most of its tools and a provider for one or two.

Five common cases

SituationUsual roleWhy
Your team uses a writing assistant or an image generatorDeployerYou use someone else’s system under your authority
You subscribe to a chatbot platform and put it on your websiteDeployer, as a ruleThe platform provides the system; you configure and use it
You build a feature on top of a model API and sell it to customersProvider of that featureYou put a system on the market under your name
HR software with an AI module screens CVs for youDeployer of a likely high-risk systemHiring is on the Annex III list
You take a vendor’s high-risk system, rebrand it or change what it is forPossibly providerArticle 25 can move provider duties to you

The website chatbot is the case that surprises people. Most of the time you are a deployer of the platform. If you go further, train it, wrap it in your own product and sell that product to others, the picture changes. That is exactly the sort of line the triage has to draw in writing.

What each role has to do

Both roles, since 2 February 2025: Article 4 asks providers and deployers to take measures so that the people working with AI have a sufficient level of AI literacy. In practice: documented measures, not certificates.

Transparency, Article 50, since 2 August 2026:

  • Providers design systems that interact with people so those people know they are talking to an AI, and mark synthetic output in a machine-readable way (with extra time until 2 December 2026 for generative AI already on the market).
  • Deployers disclose deepfakes, disclose AI-generated text published on matters of public interest unless a human holds editorial responsibility, and inform people exposed to emotion recognition or biometric categorisation.

High-risk systems, from 2 December 2027 for the Annex III list:

  • Providers carry the heavy load: risk management, data governance, technical documentation, logging, human oversight design, conformity assessment and registration.
  • Deployers use the system according to its instructions, assign human oversight to people with the competence to do it, monitor it, keep the logs it generates (at least six months, unless other law says otherwise), and inform workers before using it in the workplace.

How to work it out for your company

  1. Make the list. Every AI system in use, including the ones that arrived inside software you already pay for, and the ones staff use without asking.
  2. For each system, ask two questions. Who put it on the market, and under whose name? What do we use it for, and does that use appear in Annex III?
  3. Write the reasoning down. A one-line justification per system is worth more than a confident answer nobody can reconstruct.
  4. Check your contracts. If you are a deployer of something high-risk, you will need the provider’s instructions and documentation. Ask now, not in 2027.

This is the core of the first deliverable in our AI Act Readiness Review: an inventory, then provider or deployer and a risk category for each system, with the reasoning written down.

Sources. Regulation (EU) 2024/1689, Articles 3(3), 3(4), 4, 25, 26 and 50, as amended by Regulation (EU) 2026/1744 (Digital Omnibus on AI). Dates verified on 5 October 2026.


A practical readiness and implementation engagement. It does not replace your legal counsel, a formal conformity assessment or certification. We work alongside your counsel, never in its place.