Ethics without teeth? Challenges and opportunities in AI declarations for platform governance

ABS 1 International audience The rapid integration of artificial intelligence (AI) into digital platforms has raised critical questions about how AI’s ethical declarations influence this sector. This study adopts a mixed-methods approach. First, a descriptive content analysis examined 54 declarations, including 45 national declarations across Africa, Asia, Europe, and the Americas, and 9 from major global actors (MGAs) such as the OECD, G7, and the EU. Ethical principle frequency was examined, and a benchmarking index was developed to compare “dominant principles” cited in over 50% of regional declarations with those cited in over 50% of MGA declarations. The analysis reveals universal adoption of societal well-being, fairness, accountability, and privacy (100%), while transparency and security show regional variation (75%). Second, a semi-systematic literature review following PRISMA guidelines identified four opportunities (e.g., global participation) and seven limitations (e.g., lack of standard frameworks, definitional ambiguities, implementation challenges, and legal enforcement difficulties). The implications of these limitations for digital platforms are then examined, leading to the identification of two dimensions for responsible platform governance: assessment mechanisms (e.g., UNESCO’s Ethical Impact Assessment) and governance implementation structures. The study further distinguishes three tiers of enforceability: declarative, procedural, and institutionalized ethics, bridging normative declarations and operational practice in platform governance.