Executive summary
The EU AI Act, formally Regulation (EU) 2024/1689, creates a common framework to support safe, transparent and trustworthy AI. It does not treat every use case in the same way: the higher the potential impact on people, safety or fundamental rights, the stronger the requirements.
Inside an organisation, the first topics are very practical: inventory AI tools, classify use cases, check suppliers, protect data, ensure human validation and train users.
The 13 chapters of the AI Act in summary
General provisions
This chapter defines the purpose of the Regulation, its scope, key definitions and the AI literacy obligation. It creates the common vocabulary: AI system, provider, deployer, importer, distributor, data, risks and responsibilities.
Prohibited AI practices
The Regulation bans uses considered to create unacceptable risk, such as certain harmful manipulation, exploitation of vulnerabilities, social scoring and some abusive uses linked to biometrics and surveillance.
High-risk AI systems
This is the most operational chapter. It explains how high-risk systems are classified and introduces strong requirements: risk management, data governance, technical documentation, registration, logs, transparency, human oversight, accuracy, robustness and cybersecurity.
Transparency obligations
Certain systems must clearly disclose that they use AI. Users should be able to know when they interact with AI or when content has been artificially generated or manipulated, in the situations covered by the Regulation.
General-purpose AI models
This chapter covers models that can serve many purposes, such as large language models and generative models. It introduces obligations for providers, with stronger requirements for models presenting systemic risk.
Measures in support of innovation
The Regulation introduces regulatory sandboxes, real-world testing and support measures for SMEs and start-ups. The goal is to allow innovation while keeping a control framework.
Governance
This chapter organises European and national governance: the AI Office, the European Artificial Intelligence Board, advisory forum, scientific panel and national competent authorities.
EU database for high-risk AI systems
High-risk AI systems listed in Annex III may need to be registered in an EU database. This reinforces traceability and transparency for the systems concerned.
Post-market monitoring, information sharing and market surveillance
The Regulation covers monitoring after placing systems on the market, reporting serious incidents, surveillance powers, remedies and the right to explanation for certain individual decisions.
Codes of conduct and guidelines
The Commission may publish guidelines and encourage codes of conduct, including for uses that are not strictly high-risk. These documents will help interpret and apply the Regulation.
Delegation of power and committee procedure
This chapter allows some parts of the Regulation to evolve over time. The AI Act is therefore not a frozen framework: delegated acts, adjustments and clarifications may follow.
Penalties
The Regulation includes penalties and administrative fines, with different levels depending on the seriousness of the infringement. Penalties cover prohibited practices, high-risk obligations and general-purpose AI model obligations.
Final provisions
The final chapter covers amendments to other EU texts, systems already placed on the market, evaluation, review and the progressive entry into force/application of the Regulation.
Dates to watch
1 August 2024 — the Regulation entered into force.
2 February 2025 — prohibited practices and AI literacy obligations started to apply.
2 August 2025 — governance rules and obligations for general-purpose AI models started to apply.
2 August 2026 — general application of the Regulation, with exceptions and transition provisions.
2 December 2027 and 2 August 2028 — announced timeline for certain high-risk AI rules under the Omnibus VII simplification political agreement; check the final applicable texts before relying on it.
What an IT team or SME should do now
Conclusion
The AI Act should not be read only as a regulatory constraint. It is also a useful framework to regain control over AI use cases, reduce shadow AI, secure data and clarify responsibilities. The right first step is to turn the Regulation into a simple map of use cases, risks, suppliers and priority actions.
Useful official sources
Français
This content is a practical summary for IT governance. It does not replace legal advice tailored to your context.