What Is Public and What Is Restricted in the EU Database?

What is the default public position?

The default position for Article 49 registrations is public access. Article 71(4) states that, with the exception of the section referred to in Article 49(4) and Article 60(4)(c), the information contained in the EU database registered in accordance with Article 49 is accessible and publicly available in a user-friendly manner, and should be easily navigable and machine-readable. That default serves transparency and supervision: a deployer can verify a provider's registration, a researcher can study uptake, and an authority can cross-check an entry against a system on the market. Because entries are public, they should be written for a public reader, using the exact Annex VIII items, precise identification of the system and provider, and no additional material. A provider that includes unnecessary detail in a public entry cannot later treat that detail as confidential, and a provider that omits a required item leaves a public gap that an authority or deployer can see.

Which registrations sit in the secure non-public section?

Article 49(4) provides that, for high-risk AI systems referred to in Annex III points 1, 6 and 7, in the areas of law enforcement, migration, asylum and border control management, registration is in a secure non-public section of the EU database and includes only specified information from Annex VIII Sections A, B and C and from Annex IX. Point 1 concerns biometrics, point 6 concerns law enforcement and point 7 concerns migration, asylum and border control management, and the provision applies where the system falls within those points and areas together. Only the Commission and the national authorities referred to in Article 74(8) have access to the respective restricted sections. A provider or deployer in one of these areas must therefore identify the point and area precisely, enter only the specified items, and use the secure section, rather than entering a fuller record in the public part of the database or withholding registration altogether.

For the broader regulatory framework, see EU database, registration and post-market monitoring.

Which systems are registered nationally instead?

High-risk AI systems referred to in Annex III point 2, concerning the management and operation of critical infrastructure, are registered at national level under Article 49(5), rather than in the EU database. That national route reflects the national supervision and security sensitivity of critical infrastructure systems, and it means a provider of such a system should not expect an EU database entry to satisfy its registration duty. The national registration process, the authority concerned and the information required are matters for the Member State concerned, and they should be confirmed with that authority before the system is placed on the market or put into service. A provider operating critical infrastructure systems in more than one Member State should record the national route separately for each Member State, since a registration in one Member State does not automatically satisfy the position in another.

How is testing information treated?

Information registered under Article 60 concerning testing in real world conditions is treated differently from ordinary Article 49 information. Article 71(4) states that information registered in accordance with Article 60 is accessible only to market surveillance authorities and the Commission, unless the prospective provider or provider has given consent for it also to be made accessible to the public. The testing registration under Annex IX covers the test identification, the provider and deployers involved, a brief description of the system and its intended purpose, a summary of the main characteristics of the testing plan, and information on suspension or termination. A provider planning a real-world test should therefore prepare the Annex IX information knowing it will be restricted by default, decide deliberately whether to consent to public access, and keep the testing record updated throughout the test as the provision requires.

How do you identify the right access category?

Identifying the access category requires three checks, in order. First, identify the Annex III point and area: does the system fall within points 1, 6 or 7 in law enforcement, migration, asylum or border control, or within point 2 on critical infrastructure? Second, identify the registration route: EU database, secure non-public section, or national registration? Third, identify the information set: which Annex VIII section and which specified items apply, or, for testing, which Annex IX items apply? Each check should be recorded with the provision relied on and the reasoning, because an error at the first check produces an entry in the wrong place with the wrong access. Where a system spans more than one point or area, the category should be confirmed with the relevant authority before submission, and the confirmation should be kept with the registration record.

Continue with what the EU database contains and who can access it.

What does this guide not decide?

This guide does not assign any particular system to the public, restricted or national route, and it does not reproduce the full item-by-item content of the restricted sections, which is limited by the provision itself to specified Annex items. Access classification depends on the system's Annex III point, its area of use and the operator role, and it requires the consolidated text of Articles 49 and 71 and Annexes VIII and IX to be checked for the specific case. The guide also does not describe national registration procedures, which vary by Member State. Where access classification is uncertain, or where a system touches a sensitive area, confirm the category with a competent authority or legal adviser before submitting any entry.

Frequently asked questions

Continue through this AI Act cluster

• EU database, registration and post-market monitoring
• what the EU database contains and who can access it
• testing in real world conditions under Article 60
• which deployers must register their use
• law enforcement AI systems
• critical infrastructure safety components