The Act is organised so that scope and definitions come first, prohibited practices and classification follow, high-risk requirements and actor duties form the centre, general-purpose models, transparency, innovation support, governance and enforcement follow, and application dates close the text. For a single system, the efficient reading order is Article 2 scope, Article 3 definitions, Article 5 prohibitions, Article 6 classification, the actor duties in Chapter III Section 3, Article 50 transparency, and Article 113 application dates, with the relevant annex checked at the classification step and the consolidated text used throughout.
What does Chapter I establish?
Chapter I contains the subject matter, scope, definitions and AI literacy provisions. Article 1 states the purpose of the Regulation. Article 2 sets scope and territorial reach. Article 3 defines the terms used throughout the Act. Article 4 requires providers and deployers to take measures to ensure a sufficient level of AI literacy for staff and others dealing with the operation and use of AI systems, taking account of technical knowledge, experience, education and training and the context of use. Reading Chapter I first prevents the common error of debating a duty before establishing whether the Act reaches the activity and which defined role the organisation holds.
What do Chapters II and III do?
Chapter II contains Article 5, the prohibited AI practices. It is short and decisive: a practice that matches Article 5 is prohibited, subject to the exceptions and conditions stated in that article. Chapter III is the largest chapter. Section 1 classifies high-risk systems through Article 6 and the annexes. Section 2 sets the requirements for high-risk systems in Articles 8 to 15, covering risk management, data governance, technical documentation, record keeping, transparency to deployers, human oversight, and accuracy, robustness and cybersecurity. Section 3 sets actor duties in Articles 16 to 28, including providers, authorised representatives, importers, distributors, deployers, fundamental rights impact assessment, and value-chain responsibilities. Section 4 addresses notified bodies, and Section 5 addresses standards, conformity assessment, declarations and CE marking.
For the broader regulatory framework, see EU AI Act fundamentals.
What do Chapters IV to VI cover?
Chapter IV contains Article 50, the transparency obligations for providers and deployers of certain AI systems, including interaction disclosure, synthetic content marking, and deepfake and text disclosure duties. Chapter V addresses general-purpose AI models, including classification of models with systemic risk, provider obligations, authorised representatives for model providers, and related duties. Chapter VI addresses measures in support of innovation, including AI regulatory sandboxes, real-world testing, and support for small and medium enterprises and start-ups. A reader who only reads Chapter III will miss the transparency and model duties that can apply to a system that is not high-risk, so Chapters IV and V should be checked even after a not high-risk conclusion.
What do Chapters VII to XIII add?
Chapter VII addresses governance, including the AI Office, the European Artificial Intelligence Board, the advisory forum, the scientific panel, and national competent authorities. Chapter VIII addresses the EU database for high-risk systems, registration, and post-market monitoring and information sharing. Chapter IX addresses market surveillance, enforcement, penalties and related procedures. Chapters X to XIII address codes of conduct, delegation and committee procedures, confidentiality, amendments to other Union legislation, and final provisions including Article 113 on application dates and Article 111 on transitional arrangements. Those later chapters matter for planning, because they set who enforces the Act, how registration and monitoring work, and when each provision applies.
What is the best reading order for one system?
For one system, read in decision order rather than in chapter order. Confirm scope under Article 2 and the definitions under Article 3. Screen Article 5. Classify under Article 6 with Annex I and Annex III open beside the text. Identify the actor role and read the matching duties in Chapter III Section 3. Check Article 50 and, where a model is involved, Chapter V. Then check Article 113 for the application date of each provision relied on, and Article 111 for any transitional provision. Keep the consolidated text open throughout, because Regulation (EU) 2026/1744 amended several articles and dates, and an unamended 2024 text will give wrong dates for Annex III and Annex I high-risk application and will omit the December 2026 prohibitions.
Continue with EU AI Act timeline from 2024 to 2028.
What can a chapters map not do?
A map shows where provisions sit and how they connect. It does not interpret any provision for a specific system, and it does not replace reading the article that governs the question. Recitals can help explain purpose, but the operative articles and annexes decide the position. This guide therefore does not determine scope, classification or duties for any reader's system. It gives the route through the text so that the relevant article can be found, read in full, and confirmed with a competent authority or legal adviser where the reading is contested or the stakes are high.
Frequently asked questions
Should the Act be read from Article 1 to the end?
For understanding the structure, yes. For deciding a question about one system, decision order is more efficient: scope, definitions, prohibitions, classification, actor duties, transparency, and application dates, with the annexes checked at the classification step.
Why must the consolidated text be used?
Regulation (EU) 2026/1744 amended Regulation (EU) 2024/1689, including application dates in Article 113, Article 5 prohibitions, and Annex I. The original 2024 text alone will give outdated dates and omit new provisions.
Where are the application dates found?
Article 113 sets the application dates for the provisions of the Act, as amended by Regulation (EU) 2026/1744. Article 111 sets transitional provisions for systems and models already placed on the market. Both should be checked for the specific provision in question.
Continue through this AI Act cluster
This article is an informational overview, not a legal determination. Check the current legal text and official implementation guidance for your system and role.
Consolidated Regulation (EU) 2024/1689 ↗AI Act Service Desk timeline ↗European Commission regulatory framework for AI ↗