Artificial intelligence in anti-corruption: comparative insights from public procurement across the EU

This thesis explores Artificial Intelligence-based Anti-Corruption Technologies (AI-based ACTs) in the public sector. Amid demands for innovative anti-corruption strategies and government digital transformation, AI holds promise to help counter corruption, but also introduces new challenges and risks. Despite relevance, existing studies provide limited insight into how and why public sector actors design and adopt AI-based ACTs, and what implications arise. This research aims to explore how, why, and with what implications public bodies engage with AI in anti-corruption. It focuses on public procurement within the EU, characterised by vast data flows and vulnerability to corruption. Bridging anti-corruption innovation and public-sector AI, the research draws on interdisciplinary foundations. It uses a qualitative, exploratory, and comparative case study design across four EU countries – Italy, Germany, Estonia, and Cyprus –, selected to capture diverse contexts in perceived corruption, digital government maturity, and public procurement openness. Data collection triangulated multiple sources: interviews with public officials, data scientists, IT developers, and subject-matter experts; desk research of official and unofficial documents, and elements of participant observation. Data analysis combined thematic analysis, process tracing, and cross-case comparison. The research produced several key findings. First, a distinctive interplay of institutional and sociocultural factors shapes each country’s engagement with AI-based ACTs. Second, these applications in public procurement are emergent and vary by anti-corruption role and sustainability horizon. Third, engagement with AI-based ACTs reflects interdependent micro, meso, and macro-level factors acting as both enablers and roadblocks, with country-specific patterns. Lastly, AI-based ACTs in the public sector are not neutral tools but socio-technical assemblages shaped by symbolic meanings, strategic goals, procedural arrangements, gendered configurations, and ethical concerns. By advancing empirical, theoretical, and practical understanding of both technology-based anti-corruption and public-sector AI, this research offers insights for scholars, policymakers, and practitioners.