Conformity assessment

The process by which a high-risk AI system is shown to meet EU AI Act requirements (risk management, data governance, documentation, human oversight, accuracy); harmonized standards such as ISO 42001 can support it but do not guarantee it. For most Annex III high-risk systems, providers may self-assess (internal conformity assessment, Annex VI procedures); for systems involving remote biometric identification, a notified-body third-party assessment is mandatory (Regulation (EU) 2024/1689, Article 43).

The process by which a provider demonstrates that a high-risk AI system meets the applicable requirements of the EU AI Act, as specified in Article 43 and Annexes VI and VII. Depending on the system category, this may be a provider's internal assessment or may require involvement of a notified body. The result is documented in the EU declaration of conformity under Article 47.

The procedure (Article 43) by which a high-risk system is checked against the Act's requirements before market. It runs by internal self-assessment (Annex VI) for most Annex III systems, or by an independent notified body (Annex VII) for biometric systems where harmonized standards are not applied.

The process of demonstrating that a high-risk AI system meets the requirements of Chapter III, Section 2 (Article 3(20)); the expert prepares for and coordinates it but does not conduct it, since that is the notified body's or the provider's self-assessment role depending on the route.

The process for demonstrating whether the requirements of the EU AI Act applicable to a high-risk AI system have been fulfilled. Routes are specified in Article 43, ranging from internal assessment for Annex III standalone systems to third-party assessment by a notified body for certain high-risk systems.