The article discusses the issues and directions of regulation of the development and application of artificial intelligence (AI). The existing approaches to legal, ethical and technical regulation of AI at the national and international levels are analyzed, their shortcomings and limitations are revealed. A detailed analysis of various approaches to AI regulation (soft, hard, neutral, ethical, technical) is presented. The need for a balanced, integrated approach to regulation is emphasized, which will stimulate the development of AI, while minimizing risks and ensuring the protection of citizens' rights. It is shown that choosing an approach to AI regulation is a matter of balancing innovation, safety and ethics, which should take into account the interests of all parties, not just those who develop and use these technologies. It is concluded that the formation of an effective AI regulatory system is a difficult but extremely important task to ensure its safe and ethical development.