International Journal of Scientific Research and Engineering Development

Web-based systems are increasingly exposed to security threats, and traditional fixed security mechanisms are often unable to deal with new and constantly changing attack patterns. Most existing approaches depend either on predefined security rules or machine learning models, which limits their ability to identify and respond to complex threats in real time. This paper proposes a Hybrid Dynamic Security Policy Enforcer Using Artificial Intelligence that combines rule-based security policies with an unsupervised machine learning algorithm, Isolation Forest, to identify unusual activities and dynamically enforce security policies. The system continuously observes user login behavior, session activities, and network traffic, and assigns a risk score ranging from 0 to 100 for each interaction. When abnormal behavior is detected, the active session is automatically frozen, and an email verification alert is sent to the legitimate user. The system also uses an adaptive policy mechanism in which repeated violations result in increasingly strict actions, starting with a five-minute verification period for the first offense, followed by a two-minute period for the second offense, and a permanent block for the third offense. In addition, the system detects SQL injection and cross-site scripting attacks, monitors ICMP traffic, tracks IPv4 and IPv6 packets with geolocation information, and provides a live dashboard for administrators. An integrated attack simulator tests three common scenarios—brute-force, credential-stuffing, and bot-based attacks—to evaluate the system's response. The experimental results indicate that the proposed hybrid approach provides better overall security by combining anomaly detection, automated response, and adaptive policy enforcement compared with rule-based and AI-only methods.

Keerthana M, Dr. Ramya K V, Koushalya A Naik, "Hybrid Dynamic Security Policy Enforcer using AI" International Journal of Scientific Research and Engineering Development, V9(4): Page(1279-1291) July-August 2026. ISSN: 2581-7175. www.ijsred.com. Published by Scientific and Academic Research Publishing.