Adversarial Artificial Intelligence and Maturity Models In Cybersecurity: A Scope and State of The Art Review

Abstract. The use of artificial intelligence (AI) models in organizational en-vironments has substantially increased exposure to adversarial cyber threats and attacks. While technical research on adversarial AI and adversarial ma-chine learning has made some progress, there is little clarity regarding the level or structure of integration of these developments into cybersecurity ma-turity frameworks or models. This study conducts a review and scoping analysis using the PRISMA-ScR methodology to systematically identify and evaluate the literature addressing adversarial attacks on AI systems and their relationship to cybersecurity ma-turity models. The results show a lack of structural correlation between re-search on the maturity and sophistication of adversarial cyberattack tech-niques and organizational cybersecurity maturity models. Most studies focus on algorithmic mitigation techniques, with limited incorporation of govern-ance mechanisms, formal risk management structures, or maturity assess-ment indicators. This review identifies dominant research patterns and signif-icant conceptual and methodological gaps. The findings provide a structured evidence base for future research aimed at bridging adversarial AI robustness and cybersecurity maturity assessment in enterprise contexts.