ZERO-TRUST AI SECURITY: INTEGRATING AI INTO ZERO-TRUST ARCHITECTURES

Abstract The fast incorporation of artificial intelligence (AI) into enterprise and critical infrastructure systems has brought new security risks that are difficult to counter with conventional defense solutions. A relatively recent model, Zero-Trust Architecture (ZTA), which is based on the principles of continuous verification, least privilege access, and micro-segmentation, holds promise as a means of counteracting the changing cyber threats. Nevertheless, the inter-section between AI-centric applications and Zero-Trust security is not well explored, especially when it comes to adaptive identity authentication, contextual access control, and anomaly detection. The study introduces a new Zero-Trust AI Security Framework which incorporates AI into ZTA to augment the security of AI-based systems. The suggested model highlights three pillars, namely, AI-driven authentication via continuous user and device verification; adaptive authorization according to the contextual policies and dynamic trust rating; and AI-driven anomaly detection in order to recognize the malicious patterns and adversarial behaviors as well as insider threat in real time. The framework is documented as being scalable to cloud, edge and enterprise settings and may be used in healthcare, finance and government services. This research introduces a powerful adaptive and proactive AI-based application protection to the matter of securing AI-based applications in the next generation by integrating AI in Zero-Trust concepts, which will advance theoretical and practical work in relation to cybersecurity resilience.