A proposed standard for governing autonomous AI deployment

This paper introduces Institutional Control Architecture (ICA), a framework designed to bridge the gap between autonomous AI policy and operational reality. ICA provides an assessment methodology and certification regime to ensure autonomous systems are observable, bounded, and independently verifiable. It discusses three key bottlenecks in current deployments and defines core framework elements including five foundational laws, three non-negotiable constraints, eleven constitutional articles, and seven control layers.