Architecture-Level Safety Enforcement for Autonomous Systems through Non-Bypassable Execution Control

This paper introduces an architecture-level approach to safety enforcement for autonomous systems based on non-bypassable execution control. It positions safety not as a behavioral or policy-driven property, but as an intrinsic system property enforced through architecture. The proposed framework establishes a strict separation between decision generation and execution authorization, ensuring that all actions are validated before execution. Rather than relying on behavioral correctness or post-hoc monitoring, the architecture enforces capability gating and runtime safety validation as core system mechanisms. Candidate actions must satisfy predefined safety constraints and system-level contracts, ensuring that only admissible actions can be executed. The framework supports distributed and heterogeneous autonomous systems, including multi-agent configurations. It enables coordinated safety enforcement across system boundaries, reducing the risk of unsafe emergent behavior. This work contributes to the emerging field of architectural safety enforcement and provides a foundation for scalable, verifiable, and enforceable safety in real-world autonomous systems. Related works: - Safety-Bounded Autonomy: Architectural Safety Enforcement for Distributed Autonomous Systems- A Governance Architecture for Safe and Bounded Autonomous Systems- Non-Bypassable Execution Control in Autonomous Systems- Capability-Gated Autonomy in Distributed Autonomous Systems