Ensuring cybersecurity in the AI era: problems, methods, and institutionalization

The article describes state of cybersecurity as a field of science. Cybersecurity is defined as a set of system programming technologies that aim at finding program errors and vulnerabilities, removing them, and neutralizing the effects of the ones that remain. A brief review of required scientific methods and their implementation in Russian products is provided. Special attention is paid to the changes caused by the expansion of artificial intelligence (AI). New types of errors in and threats from the artificial intelligence systems have primarily emerged due to AI’s datacentric nature. A model for long-term development of trusted technologies is proposed, based on ideas of collaborative economy. The model has been successfully applied in various Russian Academy of Sciences’ Institute for System Programming projects implemented in two distributed centers of excellence: one for analysis of system software and another for trusted AI.