Open Collections

Abstract

Artificial intelligence (AI) has rapidly evolved from narrow, task-specific systems into complex technologies embedded in social, political, and economic infrastructures. As AI expands, its definition has shifted from a purely technical concept to a deeply sociopolitical and geopolitical one, raising critical concerns related to transparency, accountability, algorithmic bias, privacy, surveillance, and its broader impact on democracy, government communication, and freedom of expression. However, these risks are not distributed equally across the globe. Structural inequalities, resource disparities, and geopolitical differences have contributed to the emergence of an “AI divide” and a broader digital divide, disproportionately affecting countries outside the Global North. This thesis examines these inequalities through a comparative analysis of AI governance in the European Union (EU) and Turkey. While the EU has established a comprehensive, rights-based regulatory framework grounded in democratic accountability and fundamental rights, Turkey represents a hybrid case as an emerging economy, EU candidate country, and NATO member operating under different institutional and political conditions. Using qualitative comparative analysis of policy documents, regulatory frameworks, and governance mechanisms, this study investigates how AI governance is shaped across these contexts. It focuses on three key areas: ethical and regulatory frameworks, transparency and oversight mechanisms, and the implications of AI governance for democratic accountability and freedom of expression. The findings demonstrate that AI governance models are not universally transferable and are deeply shaped by geopolitical and institutional contexts. While the EU prioritizes risk-based regulation, independent oversight, and rights protections, Turkey relies on centralized, state-aligned institutions, where AI technologies can be used to reinforce existing power structures. These differences highlight how Global North-centric regulatory models may not function effectively in different contexts and can exacerbate existing inequalities. This thesis argues that the absence of a universal AI governance framework, combined with fragmented global approaches, reinforces power asymmetries and leaves countries like Turkey more vulnerable to the risks of AI. It concludes that more inclusive, context-sensitive, and globally equitable approaches to AI governance are necessary to address the growing AI divide.