The aim of the paper is to analyze theory and practice of GenAI and LLM applications in network and cloud security. The scientific exploration of generative AI in network security consistently reveals its dual potential. While it significantly enhances defensive capabilities, it also presents a powerful tool that can be leveraged by attackers. This creates a
continuous cycle of innovation, often referred to as a cybersecurity "arms race", where advancements in defensive AI are met by increasingly sophisticated AI-driven attacks