The AI Act implementation timeline means that the individual parts of the European regulation on artificial intelligence take effect gradually, not all at once. The first to apply were the provisions on prohibited practices and the requirement for an adequate level of staff knowledge, followed by the obligations for general-purpose models, and finally the most extensive requirements for high-risk systems, for which longer transitional periods are set. For companies, this implies a practical sequence: first determine whether any system in use falls among the prohibited practices, then ensure staff literacy and keep an overview of the systems in use, and only after that address the more demanding obligations according to classification. It is worth checking the specific dates in the current version of the text, since the interpretation and accompanying guidance are continually being added to.
See also: AI risk classification, Prohibited AI practices, Register of AI systems.
