Different approaches to AI governance: a comparative analysis of the UK AI security institute and the EU AI office

Dabić Dragana (Institut za međunarodnu politiku i privredu, Beograd), ddabic@diplomacy.bg.ac.rs

This paper examines the United Kingdom’s AI Security Institute and the European Union’s AI Office as two prominent institutional responses to the challenges posed by the development and deployment of artificial intelligence. Building on an analysis of their mandates, organisational structures, and operational instruments, the paper explores how these two institutions shape the understanding of risks, public policy priorities, and practical mechanisms for overseeing advanced AI systems and models. A comparative approach shows that the UK AI Security Institute operates primarily as an operational body focused on technical evaluation and testing in close cooperation with the industry. On the other hand, the EU AI Office plays a central role in implementing the regulatory framework and ensuring compliance with established rules. Particular attention is given to the implications of these differences for risk governance, innovation, and the protection of societal values. The analysis of these institutions contributes to a broader understanding of contemporary models of AI governance, which are still in the early stages of development. The paper concludes that effective AI governance requires a combination of technical expertise and regulatory structuring, thereby opening space for further institutional convergence and enhanced international cooperation.

Keywords: UK AI Security Institute, EU AI Office, Artificial Intelligence Act, regulation, governance, safety