The research examines one of the most important issues arising at the intersection of artificial intelligence (AI) and civil law – a question that is still predominantly theoretical today, but is expected to become highly practical in the near future: whether it is necessary and possible to recognise AI as an autonomous legal person, and what other legal-technical solutions might address the liability and legal certainty challenges generated by autonomous systems.

The first main part of the thesis reviews the conceptual and technological foundations of AI: it presents the development of artificial intelligence from the industrial revolutions and classical milestones of the history of science (Turing, neural networks, machine learning, generative AI), and then systematises definitions of AI from general, technical and legal perspectives. A separate chapter examines how the concept of AI has evolved during the European Union’s legislative process, and how the currently applicable AI Act definition fits with the terminology of the OECD and other international organisations. The author provides a detailed discussion of the typology of AI systems (weak/strong AI, levels of consciousness, risk-based categorisation), which later serves as the conceptual basis for comparing models of legal personality and liability.

The second part analyses the international, EU and Hungarian regulatory environment applicable to AI. The thesis surveys the framework documents of the OECD, the UN, UNESCO and the Council of Europe, and then presents in detail the European Union’s regulatory path on AI, from fundamental rights (the Charter of Fundamental Rights, data protection, the GDPR) to the AI Act adopted in 2024.

The third and central part examines the civil-law status of AI: the thesis places side by side the approaches that conceptualise AI as software, as a “thing”, or as a potential legal person, reviews the dogmatic foundations of legal personality, and discusses the debate on whether legal personality is a closed or an expandable category. The author introduces the concept of “electronic personhood” and the model of the Grishin Act, and then develops several alternatives for addressing the legal personality of AI (hybrid legal person, indirect and direct algorithmic legal personality, dependent legal personality). Particular emphasis is placed on issues of liability and damages: who bears liability, and under which liability construction, for damage caused by autonomous systems, and how risk can be allocated among developers, operators, users and any potential new “algorithmic” legal persons. In its concluding remarks, the thesis argues that future regulation must both preserve the fundamental principles of a human-centred legal order and remain open to innovative doctrinal solutions capable of addressing the new legal situations created by artificial intelligence.