STORM–Mission Assurance and System Security Engineering Methodology for Trustworthy Systems with AI

ABSTRACT

A comprehensive survey of 54,628 AI research papers over the last 5 years revealed a critical disconnect: fewer than 17% of papers address AI trustworthiness, reliability, or security. Furthermore, of the research addressing trustworthiness, 96.7% focuses exclusively on algorithmic solutions and performance. Early-life cycle mission planning and stakeholder requirements receive less than 3% of research attention. Consequently, AI systems are frequently developed and assured in a vacuum, disconnected from the operational missions they are intended to support, leaving mission assurance completely neglected. To bridge this gap, we introduce the System-Theoretic Technical and Operational Risk Management for AI (STORM-AI) methodology, an integrated approach that combats vulnerabilities in AI trustworthiness by combining two complementary workflows: System-Theoretic Process Analysis for Security (STPA-Sec) and Certified Security by Design (CSBD). STPA-Sec is a mission-first hazard analysis technique that systematically generates mission objectives, identifies unacceptable losses, and defines the necessary constraints and restraints that must govern system behavior. CSBD then translates these operational constraints into a formal
policy model, providing machine-checked formal verification that the system’s architecture enforces complete mediation. The vision of STORM-AI is establishing trustworthiness by preserving and accounting for the faithful implementation of commander’s intent at all levels in a mission system, from top-level mission descriptions down to the behavioral descriptions of hardware.

BIO

Mr. Clark, CEO of MBClark Engineering LLC, and CTO of EchoATC LLC, has 28 years of engineering experience spanning the automotive industry, the Air Force Research Lab, and the private sector (Galois). During his tenure at AFRL and Galois, Mr. Clark consulted on DARPA cyber resiliency programs such as High Assurance Cyber Military Systems (HACMS) and Cyber Assured Systems Engineering (CASE).