A Deterministic Architecture for Long-Horizon AI Systems

This record contains Version 26 (frozen preprint) of A Deterministic Architecture for Long-Horizon AI Systems.AI systems intended to operate as long-running autonomous or adaptive systems face a class of failure not captured by traditional notions of error, robustness, or performance degradation. Even when locally correct and internally consistent, such systems may become unsafe to rely upon as accumulated state and derived artifacts gradually acquire de facto authority. This failure arises not from adversarial behavior or misalignment, but from implicit authority drift driven by reuse, dependency, and ungoverned promotion of information over repeated autonomous decision cycles.We present a deterministic architecture for long-horizon autonomous and adaptive systems that prevents implicit authority drift by construction. The architecture enforces a strict separation between immutable substrate law and governed mutable state, treats authority as an explicit, versioned system property, and constrains all authority-bearing change to discrete, inspectable, and reversible governance actions. A deterministic execution pipeline with mandatory refusal semantics ensures that autonomous action selection and state transition occur only when substrate-enforced authority conditions are satisfied.The system employs a three-tier memory architecture that distinguishes ephemeral working context, persistent non-authoritative artifacts, and authoritative epistemic basis. Research, analysis, and generative components may be used freely within autonomous and adaptive workflows, but are treated as epistemically inert by default, producing artifacts rather than authority. Provenance captures not only dependency relationships but epistemic admissibility, enabling deterministic replay and post hoc validation of autonomous decisions under fixed authority configurations.This work does not aim to produce correct answers, aligned behavior, or adaptive intelligence. Instead, it presents a deterministic architectural framework that constrains how long-running autonomous systems may evolve, accumulate state, and exercise authority over time. By enforcing explicit governance, reversibility, and determinism under accumulation, the architecture supports autonomous operation in high-cost-of-failure domains where auditability, bounded authority, and correction without implicit drift are required.