A Zero-Trust Cybersecurity Architecture for AI-Enabled Smart Manufacturing Systems

Abstract

The rise of artificial intelligence (AI), IIoT, cyber-physical systems, edge, and cloud computing technologies, and intelligent automation is leading to the creation of a highly connected and data-driven space for modern manufacturing. On the other hand, it broadens the attack surface and poses challenges to traditional perimeter security models based on implicit trust inside corporate networks. In this paper, we introduce Zero-Trust cybersecurity architecture for AI-powered smart manufacturing environments. The architecture incorporates five security domains: Identity and Access Security, Device and Operational Technology (OT) Protection, Network and Communication Security, Data, Application, and AI Security, as well as Continuous Monitoring Powered by AI. The integration of those domains is realized by means of the Centralized Zero-Trust Policy and Trust Engine, which enables context-aware and adaptive access control based on the user identity, device posture, resource sensitivity, operation context, and behavioral patterns. Special emphasis is made on the dual purpose of the AI technologies as both the instrument for cybersecurity and important part of the manufacturing infrastructure that needs to be protected from manipulation and malicious activity. The study considers the issues of dynamic trust assessment, least-privilege access authorization, micro-segmentation, and continuous verification as the ways of minimizing the risks of unauthorized access and lateral movement across interconnected manufacturing resources. We also discuss practical challenges related to legacy OT equipment, real-time requirements, interoperability, scalability, and manufacturing availability. The proposed architecture offers the framework for the implementation of the Zero-Trust approach in the case of cybersecurity of AI-powered manufacturing.