Enhancing Cyber Risk Management with AI Coaching: Addressing Honey Pot Method Limitations

The contemporary cybersecurity landscape is defined by a dual challenge: the escalating complexity of cyberattacks that weaponize legitimate remote support tools, rendering traditional defenses less effective, and the persistent role of human error as a primary factor in security breaches. Conventional deception technologies, particularly honeypots, are often static and detectable, while human security officers are susceptible to cognitive fatigue and knowledge gaps. This creates a vulnerability at the intersection of technology and human factors. This paper simulates and evaluates an AI coaching model designed to enhance the capabilities of a human cybersecurity officer. Rather than merely providing answers, the AI coach functions as an adaptive training partner, delivering real-time interventions to strengthen the officer’s decision-making heuristics. The model is formally specified and simulated using Network-Oriented Modeling, which captures higher-order adaptive learning processes within the human agent. Simulation experiments and a formal risk assessment are used to evaluate the model. The results show that the AI coach increases the probability of a successful long-term response, mitigating the officer’s inherent learning deficiencies through adaptive real-time communication.