Who is a Provider?

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 has an AI system or general-purpose AI model developed, with a view to placing it on the market or putting it into service under its own name or trademark, whether for payment or free of charge.

Examples: An AI software company building a CV-screening tool. A medical device manufacturer embedding diagnostic AI in their product. A tech company releasing a foundation model API.

Who is a Deployer?

A deployer is a natural or legal person, public authority, agency, or other body using an AI system under its authority, except where the AI system is used in the course of a personal non-professional activity.

Examples: A bank using a third-party credit scoring AI. A hospital deploying a diagnostic AI system built by a medical device company. An employer using an HR platform's AI recruitment feature.

You can be both simultaneously. A company that builds a high-risk AI system and uses it internally is both a provider (they built it) and a deployer (they use it). All provider and deployer obligations apply.

Provider Obligations for High-Risk AI (Article 16)

Obligation Article What It Requires
Quality management system Art. 17 Written policies and procedures for AI Act compliance, updated throughout the system's lifecycle
Technical documentation Art. 11, Annex IV Comprehensive documentation of the system before market placement
Record-keeping Art. 12 Automatic logging of events; logs retained for specified periods
Transparency to deployers Art. 13 Instructions for use enabling compliant deployment
Human oversight design Art. 14 Technical measures enabling effective human oversight
Accuracy and robustness Art. 15 Appropriate accuracy levels; resilience against errors and adversarial inputs
Conformity assessment Arts. 43–44 Self-assessment or notified body assessment before market placement
EU declaration of conformity Art. 47 Formal declaration that the system meets AI Act requirements
CE marking Art. 48 Affix CE marking to high-risk AI systems
EU database registration Art. 49 Register the system in the EU AI database before placing on market
Post-market monitoring Art. 72 Proactive monitoring plan; serious incident reporting
Serious incident reporting Art. 73 Report serious incidents or malfunctions to national authorities without undue delay

Deployer Obligations for High-Risk AI (Article 26)

Obligation Article What It Requires
Use per instructions Art. 26(1) Use the AI system in accordance with the provider's instructions for use
Human oversight Art. 26(2) Implement the human oversight measures specified by the provider; designate qualified persons for oversight
Monitor performance Art. 26(5) Monitor for risks arising from use in specific deployment context
Report incidents Art. 26(5) Report serious incidents or malfunctions to the provider and (where required) national authorities
Log retention Art. 26(6) Keep logs generated by the AI system that are under deployer control; minimum 6 months where no specific law applies
GDPR compliance Art. 26(8) Where personal data is processed, comply with GDPR in addition to AI Act obligations
Fundamental Rights Impact Assessment Art. 27 Required for public bodies and certain private deployers using AI in Annex III areas 5–8; register assessment in EU database
Notify affected workers Art. 26(7) Where AI systems affect employees, inform workers' representatives and the workers themselves

When Deployer Becomes Provider

A deployer that modifies an existing high-risk AI system beyond the scope of the provider's intended use — or that makes substantial modifications to the system — takes on provider obligations for the modified system.

Similarly, where a provider is established outside the EU, their EU-based importer or authorised representative takes on specified provider-equivalent obligations.


Obligations for Other Actors

Importers (Article 23)

EU-based importers of high-risk AI systems from outside the EU must verify that the provider has conducted a conformity assessment, that technical documentation exists, and that the system bears CE marking. They must not place non-compliant systems on the EU market.

Distributors (Article 24)

Distributors must verify CE marking and technical documentation are in place before making systems available. They must inform providers and importers of any identified non-compliance.

Authorised Representatives (Article 22)

Non-EU providers of high-risk AI must designate an EU-based authorised representative before placing their systems on the EU market. The representative acts as the contact point for national authorities.

Informational purposes only. Nothing on this site constitutes legal advice. Role determination and obligation scope are fact-specific. Always consult qualified legal counsel. Not affiliated with the European Union or any EU institution.