ENTERPRISE-WIDE ZERO-TRUST AUTOMATION USING AI AND CLOUD-NATIVE MICROSERVICES

The rapid adoption of cloud computing, distributed applications, APIs, and microservices has expanded the enterprise attack surface and exposed limitations in conventional perimeter-based and static access-control mechanisms. This research proposes a Risk-Adaptive Zero-Trust Automation (RA-ZTA) framework that integrates artificial intelligence, continuous identity verification, dynamic policy enforcement, cloud-native microservices, continuous monitoring, and automated security remediation into a unified enterprise-wide security architecture. The proposed framework treats every user, device, application, API, and workload as potentially untrusted and continuously evaluates access requests using contextual attributes such as identity, device posture, network characteristics, behavioral patterns, workload activity, and historical security events. An AI-based risk assessment engine generates dynamic risk scores that are supplied to a policy and decision engine for adaptive authorization. The framework is implemented using cloud-native principles, containerized microservices, service-mesh security, mutual TLS, Policy-as-Code, real-time telemetry, and automated response mechanisms. The experimental evaluation compares the proposed approach with conventional RBAC, Static Zero Trust, ML-Based Zero Trust, and AI-enabled microservice security approaches using detection, response, and scalability metrics.
The experimental results demonstrate that the proposed RA-ZTA framework achieves 97.26% accuracy, 96.87% precision, 96.21% recall, 96.54% F1-score, and 0.982 ROC-AUC. Detection latency is reduced to 68 ms, policy-decision latency to 43 ms, and automated response time to 0.39 seconds, while the false-positive rate decreases to 2.11% and attack-containment capability reaches 95.4%. The framework achieves a throughput of approximately 1,470 requests/s, with CPU utilization of 57.8%, memory utilization of 61.2%, and security overhead of 13.7%. Although AI inference, continuous monitoring, service-mesh protection, and dynamic policy evaluation introduce additional computational overhead, the results indicate that the security gains substantially outweigh the performance costs for enterprise cloud-native environments. The study demonstrates that integrating AI-driven adaptive risk assessment with Zero-Trust principles and microservice-level automated enforcement can provide a scalable, responsive, and intelligent approach to enterprise cybersecurity. The proposed architecture establishes a foundation for future research involving federated risk intelligence, explainable AI, autonomous policy optimization, and multi-cloud Zero-Trust orchestration.