This article provides an initial analysis of the EU AI Act's approach to general-purpose artificial intelligence, arguing that the regulation marks a significant shift from reactive to proactive AI governance, while concerns about its enforceability, democratic legitimacy and future-proofing remain.