Evaluating Compliance in the European Union's Digital Services Act

Abstract

This dissertation examines how compliance with the European Union’s Digital Services Act (DSA) can be understood, measured, and evaluated. The DSA reshapes the governance of online platforms through procedural duties, systemic-risk mitigation, transparency requirements, and independent auditing. Evaluating those obligations requires interdisciplinary inquiry that connects legal reasoning with technical systems, organisational practices, and user experience.

Part I develops the conceptual foundations. It analyses the DSA’s regulatory architecture, the content-moderation processes affected by the Regulation, and the socio-technical complexity created by their interaction. Compliance is treated not as a static checklist but as an interpretive and organisational practice involving multiple actors, legal obligations, technical systems, and interdependent artefacts, including transparency reports, statements of reasons, systemic-risk assessments, and audit reports.

Part II develops methods for evaluating compliance. These include a consistency-based approach for comparing reporting artefacts, mixed methods for systemic-risk assessment and independent auditing, a legal codebook for analysing political and electoral content, and an examination of large language models as possible support tools in DSA audits. These approaches are intended as operational starting points for regulators, auditors, researchers, and civil-society organisations.

Part III applies these approaches in empirical studies of platform reporting, moderation decisions, user reporting pathways, AI-supported reporting, Trusted Flaggers, and systemic risks. The studies identify recurring gaps between legal requirements, platform interfaces, and the evidence made available for evaluation. The results show why claims about compliance must remain tied to the scope, data, and methods of each evaluation.

Part IV and the conclusion synthesise these findings and examine the DSA’s wider role as a governance and industry-shaping framework. They also identify external conditions that influence compliance but are not fully observable through platform reports alone, including business incentives, monetisation structures, geopolitical pressures, enforcement choices, and differences between jurisdictions. Future work should therefore test the proposed methods longitudinally and across platforms, languages, Member States, and regulatory settings, while improving secure data-access infrastructures and standards for AI-supported auditing.

Overall, the dissertation contributes concepts, methods, and empirical evidence for the evaluation of platform governance under the DSA. It argues that credible assessment depends on combining expertise from law, human--computer interaction, data science, governance studies, and AI research, and on making the assumptions, evidence, uncertainty, and limits of each compliance claim explicit.