EU AI Act Enforcement Timeline
Every enforcement phase of Regulation (EU) 2024/1689, from entry into force to the final transition deadline.
Entry into force
AI Act entry into force
Regulation (EU) 2024/1689 entered into force. The transition period began.
Phase 1-6 months after entry into force
Prohibited AI Practices
Article 5 prohibited practices are enforceable. Systems involving social scoring, subliminal manipulation, exploitation of vulnerabilities, real-time biometric identification in public spaces (with exceptions), emotion recognition in workplace/education, predictive policing based solely on profiling, and untargeted facial image scraping must be discontinued.
Phase 2-12 months after entry into force
GPAI Model Obligations
Chapter V obligations for general-purpose AI models take effect. All GPAI providers must comply with transparency, technical documentation, copyright policy, and energy consumption reporting. Systemic risk models (Article 51) face additional evaluation, incident reporting, and adversarial testing duties. The AI Office begins enforcement.
Transparency — in force since 2 August 2026
Article 50 Transparency Obligations
Article 50 transparency obligations apply. Providers and deployers of certain AI systems must inform natural persons that they are interacting with AI, and mark synthetic audio, image, video, or text content as artificially generated or manipulated, subject to the Article 50 conditions.
Article 5 Omnibus deferred prohibitions — 2 December 2026
NCII / CSAM Ban (Article 5)
Article 5(1), first subparagraph, points (ba) and (bb), and Article 5(1a) and (1b) apply from this date (Article 113(3)(a) as amended by Regulation (EU) 2026/1744). Distinct from the Article 50(2) pre-existing synthetic-content marking deadline on the same day.
Article 111(4) transitional
Pre-existing synthetic-Content systems (art 50(2))
AI systems generating synthetic content that were placed on the market before 2 August 2026 must meet Article 50(2) marking obligations by this date (Article 111(4), as added by Regulation (EU) 2026/1744).
Phase 3-24 months after entry into force
High-Risk AI Systems (Annex III)
Full compliance required for high-risk AI systems classified under Annex III (stand-alone). Providers must meet all Chapter III Section 2 requirements: risk management (Art. 9), data governance (Art. 10), technical documentation (Art. 11), logging (Art. 12), transparency (Art. 13), human oversight (Art. 14), accuracy/robustness/cybersecurity (Art. 15). Deployer obligations (Art. 26), FRIA (Art. 27), and EU database registration (Art. 49) also apply. Market surveillance authorities begin enforcement.
Phase 4-36 months after entry into force
High-Risk Products (Annex I)
Obligations extend to AI systems that are safety components of products covered by Annex I Section A Union harmonisation legislation (e.g., medical devices under MDR, machinery under the Machinery Regulation, toys, lifts, pressure equipment, vehicles). These systems must meet both AI Act and sectoral product-law requirements. Notified bodies must be prepared for integrated conformity assessment.
Review milestone
Article 112 Review #1
First Commission review under Article 112 of the AI Act.
Evaluation milestone
Article 112 Full Evaluation Report
Commission full evaluation report under Article 112 of the AI Act.
Article 111(2) transitional
High-Risk for public authorities
High-risk AI systems intended to be used by public authorities must comply by this transitional deadline (Article 111(2)).
Phase 5, Extended transition for Annex X systems
Large-Scale EU IT Systems (Annex X)
AI systems that are components of large-scale EU IT systems listed in Annex X (SIS II, VIS, Eurodac, EES, ETIAS, ECRIS-TCN) must comply with the AI Act. Article 110 grandfathering expires for systems placed on the market before August 2027 unless significantly modified. This is the final compliance deadline in the AI Act.
Article 111(2) condition
Legacy high-Risk systems (conditional)
High-risk AI systems placed on the market before 2 August 2026 are in scope of the high-risk obligations only if they are subject to significant changes in their designs (Article 111(2)). This is a condition, not a calendar deadline.