Design of Adaptive Agentic AI Supporting Cybersecurity for Ransomware Risk Management

This paper presents a computational analysis of human susceptibility to ransomware attacks using an adaptive agentic AI design approach. The study recognises key cognitive and behavioural states for both user and agentic AI. Interaction between user and AI at the knowledge level is used to simulate real-time intervention strategies, guiding individuals toward more secure responses. By applying higher-order adaptivity, the simulations reveal how AI guidance can alter behavioural trajectories, reducing vulnerability and improving cybersecurity resilience. The findings demonstrate that the considered knowledge-level AI interventions are able to prevent unwanted behavioural patterns typically observed in ransomware scenarios and support learning effects over time. This work contributes to the development of intelligent, human-centered cybersecurity strategies that adapt dynamically to behavioural risk patterns.