Get an overview over our software projects, research papers, Articles, summaries, data and many more.
We invite you to join our community, meet up at our events and connect with our experts.
A curated set of the primary legal texts, codes of practice and Commission guidelines we use in our AI Act trainings. Every link opens the original official source.
How to use this page
All documents are hosted by the European Commission and EUR-Lex, not by us. Click any card to open the official source in a new tab. Documents are in English.
The binding legal texts and the consolidated version
The full and binding text in its original version as published in the Official Journal on 12 July 2024, which entered into force on 1 August 2024. Includes Annexes I to XIII and Recitals 1 to 180, which serve as an aid to interpretation.
Amends the AI Act. Published in the Official Journal on 24 July 2026, entered into force on 27 July 2026. Adjusts high risk deadlines, adds two prohibited practices, eases the AI literacy duty, and inserts the new Annex XIV.
Merges the AI Act and the Digital Omnibus into a single text, as at 27 July 2026. Provided for information only and not legally binding.
On the practical implementation according to Article 96(1)(f) of the definition in Article 3, point (1)
Clarifies what does and does not qualify as an AI system under the AI Act. Important as a first step to check whether the AI Act applies at all. Not an AI system → the AI Act does not apply.
On the practical implementation according to Article 96(1)(b) of the prohibitions in Article 5
The Commission's reading of the Article 5 prohibitions, with examples of what falls inside and outside each ban.
On the practical implementation of Article 6 under Article 6(5) in conjunction with Article 96
Covers both classification routes: high-risk AI systems under Article 6(1) in conjunction with Annex I where existing product legislation applies, and under Article 6(2) in conjunction with Annex III. Draft, published 19 May 2026.
Guidelines on scope, definitions and exemptions for paragraphs 1 to 5, plus a code of practice on operationalisation
Clarifies the scope, definitions and exemptions of the transparency obligations under Article 50(1) to (5), with practical examples. Published 20 July 2026.
Operationalises the marking duties for AI-generated content under Article 50(2) and for deepfakes and AI-generated text of public interest under Article 50(4). Published 10 June 2026.
Chapter V, Articles 51 ff., regulates the model layer. Only providers of GPAI models are subject to the obligations.
Clarifies when a model counts as a general-purpose AI model, who qualifies as a provider, and which obligations follow. Published 18 July 2025.
Voluntary code that helps providers of general-purpose AI models show how they meet their obligations. Published 10 July 2025.
Disclaimer
This page is an internal navigation aid for our trainings and links to official EU sources only. It is not legal advice and does not reproduce the documents. Links last reviewed 11 August 2026.
