Article 3 defines the building blocks of the Act, including AI system, general-purpose AI model, provider, deployer, importer, distributor, authorised representative, placing on the market, putting into service, intended purpose, reasonably foreseeable misuse, substantial modification and safety component. Later articles use those terms in their defined sense, so a classification or duty analysis that uses an everyday meaning of provider or deployer instead of the defined meaning will go wrong even if the facts are correctly recorded.
Why do definitions carry so much weight in this Act?
The EU AI Act is a definitions-driven regulation. Article 5 prohibits practices described in defined terms, Article 6 classifies systems by reference to defined concepts, and Chapters III and V attach duties to defined roles. If the defined term is misread, the rest of the analysis inherits the error. A team that calls every supplier a provider, or every customer a deployer, will assign duties to the wrong party. A team that treats a model as a system, or a component as a finished system, will test the wrong object against Article 6. The discipline is simple to state and harder to practise: quote the definition, record the facts that match each element, and keep everyday labels out of the decision record.
What is an AI system under Article 3(1)?
Article 3(1) defines an AI system as a machine-based system designed to operate with varying levels of autonomy, that may exhibit adaptiveness after deployment, and that, for explicit or implicit objectives, infers from the input it receives how to generate outputs such as predictions, content, recommendations or decisions that can influence physical or virtual environments. Three elements deserve attention. Machine-based operation and inference separate the concept from ordinary deterministic software that only applies fixed rules without inference. Varying autonomy and possible adaptiveness describe how the system operates after deployment. The output element requires that the generated output can influence an environment. All elements must be considered together, and a system that fails one element needs its reasoning recorded, not assumed.
For the broader regulatory framework, see EU AI Act fundamentals.
What is a general-purpose AI model under Article 3(63)?
A general-purpose AI model is defined separately from an AI system. It is an AI model, including one trained with a large amount of data using self-supervision at scale, that displays significant generality and is capable of competently performing a wide range of distinct tasks regardless of how it is placed on the market, and that can be integrated into a variety of downstream systems or applications. The definition excludes models used for research, development or prototyping before they are placed on the market. The distinction matters because Chapter V imposes duties on providers of general-purpose AI models that are different from the duties on providers of high-risk AI systems, and a downstream system built on such a model still needs its own classification.
How are provider, deployer, importer and distributor defined?
A provider is a natural or legal person, public authority, agency or other body that develops an AI system or a general-purpose AI model, or that has one 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. A deployer is a natural or legal person, public authority, agency or other body using an AI system under its authority, except where the use is in the course of a personal non-professional activity. An importer is a person located or established in the Union that places on the market an AI system bearing the name or trademark of a person established outside the Union. A distributor, other than the provider or importer, makes an AI system available on the Union market. Article 25 can change which person counts as provider, so roles must be rechecked when branding, modification or purpose changes.
What do placing on the market and putting into service mean?
Placing on the market means the first making available of an AI system or a general-purpose AI model on the Union market. Putting into service means the supply of an AI system for first use directly to the deployer or for own use in the Union for its intended purpose. The difference matters for timing and for who acts first. A system supplied to a deployer for first use may be put into service without a conventional sale, and a system developed for own use can still be put into service. Records should distinguish first making available from later distribution, resale or continued use, because transitional provisions and application dates often turn on when a system was first placed on the market or put into service, not on when it was last updated.
Continue with what counts as an AI system.
What remains uncertain even after reading Article 3?
Definitions do not remove every borderline case. Intended purpose, reasonably foreseeable misuse, substantial modification and safety component all require judgement on real facts, and the Commission guidance and case practice will shape how those terms are applied. A definition can tell a reader what to look for, but it cannot by itself settle whether a particular feature infers, adapts, or materially influences an outcome. This guide therefore does not determine whether any particular technology is an AI system or a general-purpose AI model, or which role any particular organisation holds. Those determinations need the full facts, the current consolidated text, and confirmation with a competent authority or legal adviser where the answer is contested.
Frequently asked questions
Is every AI model an AI system?
No. Article 3 defines AI systems and general-purpose AI models separately. A model can be integrated into a system, and the system built on it needs its own assessment against the Act, including its intended purpose and use context.
Can one organisation be both provider and deployer?
Yes. Roles attach to activities, not to company types. An organisation can provide one system and deploy another, and it can hold both roles for the same system in different capacities, so each activity should be recorded separately.
Does free supply change the provider definition?
No. The provider definition applies whether the system or model is supplied for payment or free of charge, provided the other elements of the definition are met.
Why does intended purpose appear in so many definitions?
Intended purpose, defined as the use for which a system is intended by the provider, including the specific context and conditions of use, anchors classification, documentation and oversight. It is the purpose as specified by the provider, evidenced by instructions, promotion and technical documentation, not a later user's private hope for the system.
Continue through this AI Act cluster
This article is an informational overview, not a legal determination. Check the current legal text and official implementation guidance for your system and role.
Consolidated Regulation (EU) 2024/1689 ↗AI Act Service Desk timeline ↗European Commission regulatory framework for AI ↗