Algorithmic Risk and Regulatory Asymmetry: A Türkiye-Focused Sociological Analysis of the European Union Artificial Intelligence Act

Abstract

This study offers a sociological analysis of the European Union Artificial Intelligence Act (the AI Act) as a regulatory regime that constructs algorithmic risk as a classifiable and manageable object. Algorithmic risk is the likelihood of harm, often diffuse and unequally distributed, that the decisions of AI systems produce for individuals and social groups. The article argues that the Act’s tools of transparency, explainability, and accountability assume a subject who is competent and able to perceive, comprehend, and contest a decision yet who is often absent where algorithmic power operates most intensely. The study demonstrates that this construction of risk has transformed into a global template through the Brussels effect; however, this proliferation is not a unilateral imposition. The case of Türkiye reveals that a semi-peripheral actor has internalized the central categories not as a passive recipient but through its own agency and has also built its own regulatory institutions. Meanwhile, both the content filling these institutions and the public-sector AI infrastructure are being drawn from outside, with no AI-specific binding measures having yet been taken. This situation points to a deepening regulatory asymmetry between the center (the EU) and the semi-periphery. Bringing together critical algorithm studies, the literature on the Brussels effect, and the debates on data colonialism and the semi-periphery through the case of Türkiye, this study discusses which sociological choices the Act renders as technical necessities while making algorithmic risk manageable and how these choices produce a dynamic of alignment and divergence upon reaching the semi-periphery.