Artificial Intelligence for Cybersecurity: Challenges, Applications, and Future Trends

Nowadays artificial intelligence is one of the most widely used and influential technologies. In recent years, the digital revolution has profoundly reshaped multiple aspects of everyday life, affecting fields such as culture, economy, education, and communication. Among these, cybersecurity stands out as a critical area where AI is increasingly adopted, with the promise of addressing ever-evolving threats and challenges. However, while AI has brought significant advancements, it has also introduced new challenges. As cyber threats become more sophisticated, the application of AI in this domain has proven beneficial and raised new complexities. The research work was carried out during my internship at Sicuranext and the main objective is to experimentally test and evaluate specific applications of AI, with a focus on two areas: playbook automation using the Retrieval- Augmented Generation (RAG) technique and authentication anomaly detection using Bayesian probabilistic networks. The purpose is determining whether these AI-driven solutions could be truly effective in addressing real-world cybersecurity challenges. The thesis is structured as follows: In Chapter 1 we introduce the fundamental concepts related to artificial intelligence and machine learning, outlining their main characteristics and their growing impact in the technological field. Particular attention is paid to the specific techniques used in the course of out work, which describe both their theoretical and practical aspects. This chapter therefore provides a useful conceptual basis for understanding the subsequent developments of the work. Chapter 2 is dedicated to a general overview of some aspects of cybersecurity, with the aim of illustrating the main key concepts necessary for our purpose. Chapter 3 analyzes the state of the art of already known and applied techniques that use artificial intelligence in the field of cybersecurity. Through a critical review of the literature, the most significant approaches, the result achieved, as well as the main limitations and future opportunities to improve the effectiveness of these methodologies are explored. Chapters 4 and 5 present the case studies developed during the internship in Sicuranext. They explain the main characteristics, the methodologies adopted, the analyses performed, the objectives set, and the results obtained. This section represents the applicative heart of the thesis, highlighting how the techniques discussed in the previous chapters can be concretely implemented to address specific problems in the field. Chapter 6 draws conclusions from the topics covered in this thesis and discusses the likely future trends in the use of AI in cybersecurity. This chapter aims to summarize the key findings of the work, highlighting the strengths and limitations of the analyzed approaches. Furthermore, it provides insights into emerging trends and potential advances in AI-based cybersecurity solutions, offering a perspective on how the field may evolve in the coming years.