Abstract:
The European Union has produced the world's first comprehensive regulation on Artificial Intelligence (AI). The two main regulatory instruments mobilised by the EU are the binding 2024 AI Act and the voluntary General-Purpose AI (GPAI) Code of Practice, a combination that reflects a structural tension between the regulatory oversight exercised by EU regulators and the growing role of technology platforms as co-stewards of regulation. Drawing on the concept of sociotechnical imaginaries, understood as shared visions of desirable futures and risks to be avoided, we ask: What sociotechnical imaginaries are embedded in the EU's approach to AI? To explore this question, we empirically compare the imaginaries embedded in the 2024 AI Act, the 2025 GPAI Code of Practice and the preceding 2019 High Level Expert Group on AI report 'Ethics guidelines for trustworthy AI' as a window to trace the imaginaries upon which the EU has built its approach to AI. We examine how the EU articulates the promises and risks of AI, which actors are legitimised to govern safety, what solutions are promoted, and what type of regulatory leadership emerges. We show how responsibilities are distributed between EU institutions and Big Tech, highlight the risks of regulatory capture, technosolutionism, and reliance on private infrastructure, and point towards democratic alternatives.