Regulation (EU) 2024/1689, the EU AI Act, reached general application on 2 August 2026, completing the final phase of its phased implementation. The European Commission's AI Office and national market surveillance authorities began active enforcement on that date. The preceding phases had already brought into effect: prohibitions on unacceptable-risk AI practices under Article 5 and AI literacy obligations under Article 4, operative from 2 February 2025; and governance rules and general-purpose AI model obligations under Chapter V, operative from 2 August 2025.
Article 5 prohibits AI practices including real-time remote biometric identification in public spaces by law enforcement (with limited exceptions under Article 5(1)(h)), social scoring by public authorities, and cognitive behavioural manipulation targeting vulnerable groups. Article 43 requires conformity assessment of high-risk AI systems listed in Annex III, carried out by internal control under Article 43(2) or third-party notified body assessment under Article 43(3). Article 11 mandates technical documentation before market placement. Article 71 requires registration of high-risk AI systems in the EU-wide database. Article 50 requires deployers of interactive AI systems to inform users they are dealing with AI, and mandates machine-readable marks on AI-generated content.
Providers placing high-risk AI systems on the EU market in sectors including employment screening, biometric categorisation, critical infrastructure, law enforcement, migration, and access to essential private services bear primary compliance burdens: they must register systems, maintain technical documentation, implement post-market monitoring plans, and file serious incident reports with national market surveillance authorities. General-purpose AI model providers that supply API access or integrated model services to downstream AI application developers must deliver technical documentation to those developers and, for models carrying systemic risk under Article 51, conduct adversarial testing before each major model update. Importers and distributors must verify supplier conformity before placing systems on the EU market.
AI systems placed on the EU market before 2 August 2026 under existing contracts benefit from a transitional period to 2 August 2027 under Article 111(3), provided no significant change is made to the system's purpose or design. National market surveillance authorities vary in stated enforcement priorities, and the Commission's AI Office acts as the primary enforcement body for general-purpose AI models. Open questions remain on how Annex III high-risk classifications apply to AI embedded in regulated medical devices, machinery, or financial products, and on the interaction between the AI Act and sector-specific EU legislation.
Licentium advises AI system providers, deployers, and API users on EU AI Act obligations and cross-border AI regulatory strategy. We may advise on compliance positioning or connect you with technical assessors and notified bodies through our partner network. Work we undertake includes high-risk AI system classification, Annex III mapping, conformity assessment preparation, technical documentation review, general-purpose AI model transparency compliance, serious incident reporting procedures, and regulatory engagement with national AI authorities.