Challenges and prospects of global ethical restrictions on artificial intelligence: Lessons learned from the regulatory approaches of the EU and the UN

Abstract

This study examines the global system of ethical artificial intelligence (AI) regulation, focusing on supranational initiatives of the United Nations and the European Union, as well as AI implementation in public administration. It analyzes the challenges of applying ethical principles and prohibitions and evaluates prospects for developing an effective global regulatory architecture that minimizes risks without restricting technological innovation. The study systematizes the main ethical risks associated with AI use in public life, compares the regulatory approaches of the EU and the UN, and identifies obstacles to global consensus, including technological asymmetry, the absence of universal definitions, and limited monitoring mechanisms. The research combines a review of theoretical literature and regulatory documents, qualitative content analysis, comparative analysis, and a simulation experiment. The experiment modeled budget allocation between social assistance and business support by two agents: a rational AI algorithm and a socially oriented human. The AI allocated an average of 46.7% of resources to social assistance and achieved an economic growth index of 0.472, whereas the human allocated 62.7% to social assistance, resulting in a lower growth index of 0.389. These findings demonstrate a fundamental tension between economic optimization and social values. The results indicate that ethically significant decisions should not be fully automated. The study therefore proposes a hybrid governance model combining algorithmic efficiency, human oversight, social justice, adaptability, inclusiveness, and evidence-based regulation.