A CRITICAL ANALYSES OF CIVIL LIABILITY MODELS FOR AI-ENABLED PLATFORMS IN EUROPEAN UNION

Artificial intelligence technologies become an integral part of online platforms via recommender systems, automated content moderation, ranking, profiling, generation and other decision support or decision-making technologies. Such technologies may produce or exacerbate legally actionable harm while making it difficult to establish defects, faults and causation in traditional ways. This paper provides a critical analysis of the civil-liability framework for AI-powered platforms emerging in the European Union, in particular considering the interplay between the Artificial Intelligence Act (AI Act), the Digital Services Act (DSA), the 2024 Product Liability Directive (PLD) and the withdrawn Artificial Intelligence Liability Directive (AILD). Using doctrinal, comparative and policy analysis, this study explores how ex-ante risk regulation works in tandem with ex-post compensation, whether this interaction is adequate to account for platform-specific harms, evidentiary asymmetries and multi-actor responsibility. It appears that the EU has transitioned from a largely technology-neutral liability regime to a layered approach where AI safety responsibilities, due diligence of the platform and product liability can complement each other. At the same time, there are some gaps regarding design-based harm, recommendation, autonomy, non-material damage and multi-actor involvement. The suggested model will be based on liability for risks sensitive to duties, adequate access to technical evidence, specific presumptions, and responsibility allocation without turning each damage caused by artificial intelligence into strict liability.