Which Agency to Regulate AI? Citizens’ Trust in the Design of a European Union Regulatory Agency

How could the AI regulator be designed in the European Union? Previous studies have identified several factors that influence trust in, and acceptance of, AI tools used in public policies and services. However, we know much less about citizens’ preferences regarding the type of body that could regulate AI. The objective of this article is to analyze which attributes of a regulatory agency generate higher levels of trust. In particular, we focus on trust in different configurations of two core components of such agencies: their management board and their ethics review committee. To address this question, we conducted a conjoint experiment with a representative sample of the Spanish population (n = 1,200). Participants evaluated profiles of AI regulatory agencies that varied across several attributes of both bodies. These attributes included different configurations of member state representation, the involvement of stakeholder groups, and the binding authority of the agency. This article contributes to the debate on which social actors, governance structures, and legal competences are most capable of fostering trust in an EU-level AI regulator.