Emerging models of national competent authorities under the EU AI Act

This paper examines the emerging models of national competent authorities under the European Union's Artificial Intelligence Act (AI Act), providing novel insights into how different jurisdictions approach AI governance institutionalization. Through systematic analysis of official documents from early implementing Member States, we identify three fundamental dimensions that characterize distinct regulatory approaches. First, the organizational architecture reveals a spectrum from centralized to fragmented oversight models, reflecting different philosophies about coordinated versus distributed AI governance. Second, the institutional choices between leveraging existing regulatory bodies versus establishing new AI-specific authorities highlight contrasting approaches to building governance capacity and expertise. Third, the regulatory scope demonstrates a critical divide between horizontal and vertical oversight frameworks, with some jurisdictions pioneering hybrid solutions that attempt to balance specialized knowledge with coordinated supervision. Our methodology combines document analysis of legislative proposals, government resolutions, and administrative acts with a three-dimensional analytical framework examining degrees of centralization, institutional arrangements, and oversight models. The findings contribute to the theoretical understanding of AI governance by revealing how different jurisdictions interpret and operationalize regulatory requirements, balance competing institutional priorities, and address the complex challenge of overseeing AI systems. Furthermore, the analysis offers insights into the evolution of AI governance structures and contributes to the broader discourse on institutional design for emerging technology oversight. While our analysis is limited to early implementers and based primarily on formal documents rather than operational evidence, it provides a foundation for future research examining the effectiveness of different models, their evolution over time, and their impact on AI innovation and oversight. Future studies could benefit from comparative analyses of implementation outcomes, stakeholder perspectives, and the practical challenges of operationalizing these different governance frameworks.