ARTIFICIAL INTELLIGENCE-BASED CYBERSECURITY: THREAT DETECTION CAPABILITIES, ADVERSARIAL THREATS, AND AN INTEGRATED MODEL FOR RISK MANAGEMENT

Abstract

This article analyzes the capabilities of artificial intelligence (AI) in the realm of cybersecurity, as well as the adversarial threats inherent to it. The objective of the research is to propose a conceptual model that integrates threat detection, model security verification, human oversight, and continuous risk management. Utilizing systemic, comparative, and functional analysis, along with methods of generalization, the study examines materials from ENISA, NIST, and MITRE ATLAS. As a result, a five-layered model has been developed, encompassing asset and data inventory, AI-driven detection, adversarial testing, human oversight, and ongoing monitoring and improvement. This model is a conceptual proposal; its effectiveness requires further validation through empirical testing