Explainability and interpretability of artificial intelligence use in cybersecurity

With the rapid advancement of internet-connected systems and artificial intelligence (AI), AI models have become integral to cybersecurity, particularly in identifying and mitigating cyberattacks. While these AI-based methods are widely utilised, they often generate results that lack interpretability, posing challenges for security experts, developers, educators and others who are involved in understanding the rationale behind specific outcomes. This interpretability gap has driven the emergence of explainable AI (XAI), a growing field focused on enhancing transparency by providing insights and visualisations that make AI decisions understandable to humans. XAI enhances the interpretability of AI-generated outcomes, enabling security experts to better understand and prioritise potential threats. Moreover, it helps to improve the efficiency and accuracy of threat assessments, thereby strengthening overall cybersecurity operations. To explore XAI’s role in this cybersecurity domain, a systematic literature review was conducted to offer a comprehensive analysis of its applications in cybersecurity. Adhering to PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines, a structured search was performed on the Scopus database, identifying 565 peer-reviewed studies. After a thorough selection process, 67 relevant papers were analysed. The findings reveal various AI subsets and XAI techniques employed for cybersecurity, challenges associated with XAI applications and future research directions in this domain. This study contributes significantly to the existing body of knowledge as it deepens understanding of XAI’s role in cybersecurity. Furthermore, it offers valuable insights for advancing research and practical implementations. This work helps to bridge the gap between AI’s technical complexity and its applicability in real-world cybersecurity scenarios.