Industrial Automatic Control Systems and Controllers Annotation << Back Key Features of Using Neurocryptography in Information Security: Theory, Methodology, Analysis Akhmedova Z.Kh., Askhabov A.Kh., Medzhidov Z.U. The article is devoted to the problems of using generative neural networks to attack cryptographic protocols, as well as the application of methods for protecting protocols from attacks. The purpose of the article is to study the possible potential of generative neural networks in an attack on cryptographic protocols. The work analyzes the operating principle of generative neural networks, their main types and models. A study of the nature of attacks on cryptographic protocols and their effectiveness was conducted, using various metrics, such as attack speed and accuracy of prediction and data recovery. Particular attention is paid to the analysis of successful examples and their application to modern approaches to protecting cryptographic protocols. Based on the analysis, the article reveals modern approaches to protecting cryptographic protocols, including strengthening encryption algorithms. The authors give a generalized description of methods of protection against side-channel attacks using adaptive systems integrating machine learning. As a research task, the authors defi ned an attempt to propose ways to improve cryptographic protocols, such as dynamic encryption, obfuscation of algorithms and the implementation of post-quantum protection methods. The authors focus their work on the study of modern approaches to protecting cryptographic systems and the principles of operation of the proposed approaches. Keywords: generative neural networks, cryptographic protocols, neural network attacks, post-quantum cryptography, protection methods, machine learning. DOI: 10.25791/asu.6.2025.1591 Pp. 30-36.