FortiLayer is an AI/ML model security analysis tool designed for high-consequence applications in defense and industrial environments. It performs internal behavioral analysis of AI models, examining how individual components make decisions rather than relying solely on output-level testing. This approach exposes vulnerabilities that output-only testing cannot detect. The tool targets two primary model types: - Computer vision (CV) models used in target recognition, autonomous systems, and industrial inspection - Large language models (LLMs) used in analyst and operator workflows FortiLayer is applicable across several verticals: - Defense and ISR: addresses misclassification and spoofing in target recognition and intelligence workflows - Autonomous and robotic systems: hardens perception and decision pipelines for navigation and real-time actions - Industrial and safety applications: secures automated inspection, quality control, and worker safety systems - LLM mission support: detects prompt injection and jailbreak attempts The tool integrates into CI/CD, DevSecOps, and MLOps pipelines and is compatible with common AI frameworks. It produces output suitable for audits and compliance documentation. FortiLayer has been developed with backing from multiple U.S. government agencies including the Navy, Space Force, Army, Air Force, DTRA, DARPA, DHS, NIST, and MDA. It is awardable through DoD procurement channels including Platform One, Tradewinds, and SBIR Phase III.