EU AI Act
Regulation (EU) 2024/1689 laying down harmonised rules on artificial intelligence (Artificial Intelligence Act)
Official text (opens in a new tab)Draft – legal and case law references not yet verified
In Plain Language
The first broad EU law on AI. It sorts AI systems by risk: some practices are banned outright, "high-risk" systems (for example in hiring, credit, education or policing) must meet requirements on data, documentation, human oversight and accuracy, and certain systems such as chatbots and deepfake generators carry transparency duties. Providers of general-purpose AI models have their own obligations.
Key Dates
- Entry into force
- Prohibited practices and AI literacy apply
- General-purpose AI model rules and governance apply
- Digital Omnibus on AI (Regulation (EU) 2026/1744) enters into force, amending this Act
- Art. 50(2) marking grace period ends for synthetic-content systems placed on the market before 2 Aug 2026
- High-risk rules for Annex III systems apply (postponed from 2 Aug 2026 by the Omnibus)
- High-risk rules for AI in regulated products (Art. 6(1), Annex I) apply (postponed from 2 Aug 2027)
- AI systems used by public authorities must comply
Verify Dates and status last checked 8 November 2026. Amended by the Digital Omnibus on AI (Regulation (EU) 2026/1744, in force 27 July 2026): high-risk obligations postponed to 2 December 2027 (Annex III) and 2 August 2028 (Annex I products), new prohibited practices added, and simplifications for smaller companies. Dates below reflect the amended timeline.
Key Articles for AI
| Art. 4 | AI literacy: providers and deployers must ensure staff have sufficient AI understanding (Omnibus softened this to encouraging measures). |
|---|---|
| Art. 5 | Banned practices, e.g. manipulative techniques, social scoring, untargeted facial-image scraping. The Omnibus added bans on AI-generated non-consensual intimate imagery and child sexual abuse material. |
| Art. 6 & Annex III | When an AI system counts as high-risk; the Omnibus narrowed some Annex III cases. |
| Arts. 9–15 | Requirements for high-risk systems: risk management, data governance, documentation, transparency, human oversight, accuracy. |
| Art. 26 | Obligations of deployers (users) of high-risk systems. |
| Art. 27 | Fundamental rights impact assessment for certain deployers. |
| Art. 50 | Transparency for chatbots, emotion recognition and deepfakes; Omnibus added a grace period to 2 Dec 2026 for older synthetic-content systems. |
| Arts. 53–55 | Obligations for general-purpose AI model providers. |
| Art. 99 | Penalties. |
Use Cases Citing This Law
Related Sources
AI Act Compliance Checker / Explorer (opens in a new tab)
The official platform provides an AI Act Explorer for chapters, annexes and recitals and a Compliance Checker to help evaluate requirements for AI systems and GPAI models. Tool results do not constitute certification or a binding legal determination.
AI Act Service Desk and Single Information Platform (opens in a new tab)
Official Commission information hub offering AI Act guidance, interactive tools and a Service Desk for questions. An information service, not a law or certification.
Voluntary initiative supporting preparation for the AI Act through community exchange and company pledges. It is not a binding legal instrument.
CEN-CENELEC JTC 21 — European AI standardisation (opens in a new tab)
European standards-development programme supporting AI governance and EU legislation. JTC 21 is a technical committee, not an adopted standard; this page does not establish that a particular standard is harmonised or cited in the EU Official Journal.
Code of Practice on Transparency of AI-Generated Content (opens in a new tab)
Voluntary code offering practical measures for marking, detection and labelling of AI-generated content under Article 50. The Commission page identifies the final code and confirmation of its adequacy.
Commission guidelines on Article 50 transparency (opens in a new tab)
Guidelines on transparency obligations for providers and deployers of certain AI systems, explaining Article 50 duties. Distinct from the voluntary code on AI-generated content.
Commission guidelines on GPAI obligations (opens in a new tab)
Guidelines for providers of general-purpose AI models explaining which providers are in scope and how their AI Act obligations apply. Interpretive guidance, not a separate law.
Commission guidelines on high-risk classification — draft (opens in a new tab)
Draft Commission guidelines on the classification of high-risk AI systems under Article 6, including practical examples. The linked institutional page describes a draft consultation document; this entry does not claim a final version has been adopted.
Commission guidelines on prohibited AI practices (opens in a new tab)
Non-binding Commission guidance interpreting the prohibited AI practices in Article 5, with legal explanations and practical examples.
EN 18286:2026 — Quality management system for EU AI Act regulatory purposes (opens in a new tab)
European standard translating Article 17 quality-management requirements into practical, auditable requirements for providers of high-risk AI systems. CEN-CENELEC reports its citation in the EU Official Journal as a harmonised standard in October 2026. Presumption of conformity concerns the corresponding provisions, not blanket compliance with the entire AI Act; the applicable Official Journal citation and scope require legal review.
GPAI Code of Practice (opens in a new tab)
Final voluntary General-Purpose AI Code of Practice supporting model providers with AI Act transparency, copyright and safety/security obligations. The code does not replace the Act.
Regulation (EU) 2026/1744 – Digital Omnibus on AI (official text) (opens in a new tab)
Official Journal text of the Digital Omnibus on AI, amending the EU AI Act: postponed high-risk deadlines, new prohibited practices, and simplification measures.