Artificial intelligence “fault” for “error” and the problem of nuclear military security in the context of a legal topicality

The discussion in the UN Working Group on the draft convention on the use of information and communication technologies for military and political purposes has recently given some hope for a compromise, but now it is obvious that the positions of the main participants - the United States and Russia - are irreconcilable. Questions about the cybersecurity of military systems are particularly delicate. Legal science can offer non-standard methods to resolve this conflict situation. Among the possible tools is the theory of security measures as a legal institution that includes security sanctions, developed by the well-known criminal law specialist Professor N. V. Shchedrin. Recently, the issue of using “artificial intelligence” for various applied purposes has become topical (and controversial). Many myths have arisen around it. For legally correct solution of arising problems in criminal, civil and other branches of law, demystification of “artificial intelligence” is required. Given the lack of information about the real use of “artificial intelligence”, the systems of control of nuclear weapons of a probable enemy, which include the complexes of automatic control of retaliatory nuclear weapons of the USSR/Russia and the United States, have become a source of increased concern. The article proposes a non-standard legal approach to drafting a convention that would reduce the levels of risk and mistrust of “intelligent” nuclear weapons control systems.