EU AI Act Officially Published With Staggered Compliance Deadlines
The European Union officially publishes its landmark AI Act, setting a phased timeline that begins with a ban on certain systems in early 2025. The regulation also carries extraterritorial reach, potentially subjecting global companies to its strict rules if their AI outputs are used within the EU.
The European Union officially publishes the EU AI Act in the Official Journal, establishing a finalized legal framework that enters into force on August 1, 2024. Rather than taking effect all at once, the law follows a staggered timetable that phases in obligations over a three-year period.
Under this schedule, prohibitions on banned AI systems take effect in February 2025, followed by rules for general purpose AI models in August 2025. The bulk of the requirements for high-risk AI systems, including those used in employment and worker management, apply starting in August 2026, while specific obligations for AI in aviation and medical devices begin in August 2027.
The regulation applies extraterritorially, meaning global companies face compliance if their AI system outputs are utilized within the EU. Organizations must evaluate their specific role, such as provider or deployer, and classify their AI systems by risk level to determine exactly which extensive technical and legal obligations apply to them.