This paper introduces the LivingAI governance architecture, a deterministic execution framework designed for autonomous intelligent systems. Unlike traditional model-centric AI architectures, LivingAI separates cognition, governance, authority, execution, and trace into independent architectural domains. The architecture introduces a governance-first execution model in which cognitive systems generate proposals but cannot directly trigger execution. All actions must pass through a deterministic governance pipeline consisting of governance routing, authority resolution, invariant validation, execution authorization, and execution. The architecture also introduces the WHY-TRACE mechanism, enabling deterministic reconstruction of decision processes and full auditability of system behavior. LivingAI is formally defined as a governed autonomous intelligent entity whose operational behavior emerges from the deterministic interaction of cognition, governance, authority, execution, and trace domains. This work establishes the LivingAI governance architecture as prior art for controlled autonomous intelligent systems.
Ondrej Škultéty (Wed,) studied this question.