FlagThis

Abstract

Perimeter-based cybersecurity defenses are increasingly inadequate as frontier AI models compress the timeline between vulnerability and exploit. The deployment of autonomous agents further complicates the threat landscape by introducing autonomy in goal interpretation and tool execution. This framework applies Zero Trust principles—never trust, always verify, and assume breach—to agentic AI services to establish secure foundational capabilities, mitigate emerging threat vectors, and implement defensive operations capable of matching the speed of autonomous threats.