The SOP project aims to develop a comprehensive framework for cyber risk assessment by integrating formal foundations, verification techniques, and quantitative, qualitative, and semi-quantitative approaches. Artificial intelligence and digital twins will be used to simulate infrastructures and threat agents. New techniques will be developed for fuzz testing, automated input generation, security verification, and vulnerability detection. The project will also propose new computational models and programming languages to support security-by-design methodologies. The UNIVR group focuses on the use of large language models (LLMs) for software vulnerability detection and on the application of logic and abstract interpretation to analyze security properties of programs, with particular interest in factors affecting analysis precision.