The Artificial Intelligence Act (AI Act) as the basis for legal regulation of artificial intelligence in the EU: review of the main provisions

This article reviews the main provisions of the Artificial Intelligence Act (AI Act), which entered into force as an EU Regulation in 2014. It is indicated that one of the main global trends in recent years is the active development of artificial intelligence and its application, and it is argued that since the AI Act is one of the first legal acts in the world designed to regulate artificial intelligence, and also taking into account Ukraine’s course towards European integration, it is important to examine the main issues regulated in the AI Act, since its provisions may also form the basis of Ukrainian legislation, the subject of regulation of which will be artificial intelligence. It is concluded that the division into types of risk that underlies the activities of a particular artificial intelligence system and is the basis for the AI Act is an extremely successful approach that separates the levels of regulation depending on the specific risk that underlies the artificial intelligence system. A brief overview of the main provisions is made and the subjects to which the AI Act applies are listed. It is concluded that the range of subjects is defined extremely broadly. The purpose of such a broad definition is clear – a wide list of subjects is necessary to prevent evasion of regulation by any means, primarily indirect ones. The very fact of using artificial intelligence systems in the EU, even if the developers and suppliers of such systems are located outside the European Union, is already sufficient for such subjects to be subject to regulation. This applies to subects located in the EU, as well as any importers, distributors, representatives. It is concluded that certain practices of using artificial intelligence are prohibited, although for some of them, such as the use of remote biometric identification systems in real time in public places for law enforcement purposes, exceptions are provided for, which are based on the principle of proportionality and provided that there is a legitimate purpose for using artificial intelligence in such a case. This allows for a flexible approach to the legal regulation of artificial intelligence – to prevent risks, while not limiting those cases where the use of such systems is appropriate and beneficial to society.