IT-industry is a strategic segment of development of any economy. One of the leading places in the development of the IT sector is occupied by artificial intelligence. Artificial intelligence has a huge potential to influence the development of any sector of the economy: industry and transportation, agriculture, oil and gas sector, etc. However, its ability to self-development allows us to talk about the potential to harm them. Therefore, the development of neural networks, the widespread introduction of artificial intelligence makes the question of the principles of its development and use extremely relevant. More and more developers think that it is necessary to solve the problems of legal and ethical restriction of artificial intelligence application at the stage of its design and creation. And for this purpose it is necessary to develop and fix at the normative level the fundamental guidelines (principles) of safe use of artificial intelligence. These principles should apply, first of all, to the safety of its use, transparency, accountability, non-discrimination and so on. The official legalization of these fundamental principles is intended to promote competition in the IT sector of the economy, to eliminate unfair competition, as well as factors of abuse of law, which will eventually serve as a necessary basis for the development of information technologies. The article raises the problem of the essential content of the Asilomar principles, as well as other principles, highlighted in science and legislation, allowing to form the basis of legal regulation of the creation and use of artificial intelligence technology in the IT-sphere. The study of law sources of different levels (international and state (national)) shows that the main principle of formation of legal regulation is the care in the sphere of individual (local) regulation of relations related to the IT-sphere. The main approach that is used in different legal systems is the approach when the legislator admits the possibility of regulating legal relations by narrowly focused acts. The specificity of such regulation consists in the fact that the accepted sources by their scope of action apply only to certain subjects. At the same time, as a rule, the norms contained in these acts are not generally binding. However, before talking about building a system of legal regulation of relations in the IT-sphere, it is necessary to formulate the principles on the basis of which the system will be developed. The conducted study of the types and content of the Asilomar principles shows that they can become a fundamental basis for the legal regulation of relations arising in the process of development and use of artificial intelligence technology. However, not all principles can form the basis of legal regulation. Some of them repeat and absorb each other (e.g., the principle of utility and universal benefit), some of them (e.g., justice) have an evaluative character and cannot fully fulfill the function of a means of legal regulation mechanism. Taking into account the specificity of the legal regulation of relations related to the use of artificial intelligence in the sphere of IT-economy, the study of local principles of legal regulation, developed by individual companies that are part of the IT-sphere (IBM, Microsoft, etc.), as well as proposed in doctrinal sources has been carried out. The necessity of using a number of these principles was substantiated. The authors formulate their own principles to be used in the creation of legal regulation of the use of artificial intelligence technology in the IT-sphere, and propose a system of these principles. The necessity of singling out as an independent category the principles of creation and use of large databases. The analysis of scientific approaches to the principle of liability allowed us to formulate our own conclusions aimed at establishing the subject of liability. It is justified that in case of causing harm by artificial intelligence it is impossible to impose liability on the person who created it (the inventor). Such an approach may stop progress. Responsibility should be borne by the person who benefits from the use of AI. It is suggested that such a person should be its owner. The necessity of compulsory insurance of AI technologies is argued. It is proved that the insurance fund should be used to compensate for damage.