A Generative Artificial Intelligence Methodology for Automated Zero-Shot Data Tagging to Support Tactical Zero Trust Architecture Implementation

A significant challenge in the implementation of the military’s cybersecurity framework for Zero Trust Architecture (ZTA) is that the current approach for data tagging is done manually, which is a time-consuming and error-prone process that undermines the efficiency and effectiveness of cybersecurity measures. This paper introduces an innovative methodology that leverages generative artificial intelligence (AI) for automated data tagging to support tactical ZTA implementation within mili- tary mission command systems (MCS). Specifically, we develop the Generative AI ZEro-trust Labeling (GAZEL) tool, which uses a fine-tuned Large Language Model combined with in-context learning for automated zero-shot tagging of MCS message data according to predefined access control categories, streamlining the path towards an agile and fortified cybersecurity posture.