Is My AI System High-Risk Under the EU AI Act?
Under the EU AI Act, an AI system is high-risk in two ways: it is a safety component of a product, or itself a product, covered by the EU laws in Annex I that need a third-party conformity assessment (Article 6(1)); or its intended use falls in one of the areas listed in Annex III (Article 6(2)). An Annex III system is not high-risk where it poses no significant risk of harm, under the conditions of Article 6(3), unless it profiles people.
The Annex III areas
Annex III lists eight areas: (1) biometrics, including remote biometric identification and emotion recognition; (2) critical infrastructure, as safety components; (3) education and vocational training, such as admission and evaluating learning outcomes; (4) employment, such as recruitment, filtering applications, promotion, termination and monitoring performance; (5) access to essential private and public services, including public benefits, creditworthiness and credit scoring (not fraud detection), risk assessment and pricing in life and health insurance, and emergency call triage; (6) law enforcement; (7) migration, asylum and border control; and (8) administration of justice and democratic processes.
The Article 6(3) exceptions
A system in an Annex III area is not high-risk where it does not pose a significant risk of harm to health, safety or fundamental rights, including by not materially influencing the outcome of decision making. That applies where it is intended to: (a) perform a narrow procedural task; (b) improve the result of a previously completed human activity; (c) detect decision-making patterns or deviations from them, without replacing or influencing the human assessment without proper human review; or (d) perform a preparatory task to an assessment.
Profiling is always high-risk
Whatever the exceptions say, an Annex III system that performs profiling of natural persons is always high-risk (Article 6(3), last subparagraph).
If you rely on an exception
A provider who considers an Annex III system not high-risk must document that assessment before placing it on the market or putting it into service, and provide it to national authorities on request (Article 6(4)).
When it applies
After the 2026 amendment to the Act, the high-risk rules apply from 2 December 2027 for the Annex III areas, and from 2 August 2028 for products covered by Annex I.
Check your use case
The EU AI Act is published on the NOMOS Exchange as an authority you can ask. Describe your system's use as facts and the Act's rules classify it, including the Article 6(3) exceptions and the profiling rule, with the article that decided. NOMOS compiled it from the official text; it has not been issued or attested by the EU, and it is not legal advice.
Example
A bank uses AI to sort incoming loan applications into folders before an underwriter reviews each one.
The team assumes any credit-related AI is high-risk, or that none of it is, and plans compliance on a guess.
Asked as facts, the rules weigh whether the system evaluates creditworthiness (Annex III, point 5(b)), whether it only performs a narrow procedural task (Article 6(3)(a)), and whether it profiles applicants, which would make it high-risk whatever else applies.
Frequently asked questions
Related concepts
Check your use case
Ask the EU AI Act on the Exchange: describe your system as facts and get its classification, with the article that decided.