Yesterday, 2 February 2025, the first provisions of the EU AI Act became applicable. These provisions relate to AI literacy (Chapter 1) and prohibited systems (Chapter 2). Subject matter, scope, and definitions, also in Chapter 2, will additionally apply, although they do not require explicit action by any AI actors. The next major provisions that will apply include those for general-purpose AI models and penalties (2 August 2025). The Act generally applies from August 2026, while provisions on high-risk AI systems in connection with Annex I references to existing EU legislation and relevant obligations start applying from August 2027. In this blog post, we outline what you need to know about the provisions on AI literacy and prohibited systems that you now need to comply with.

AI systems prohibited under the EU AI Act