Meaningful Human Control of Autonomous AI in Space: A Legal-Technical Mechanism

This paper addresses a critical challenge for legal scholars, policymakers, and space industry stakeholders: the structural gap in international space law exposed by the rapid deployment of autonomous artificial intelligence (AI) in outer space operations. Article VI of the 1967 Outer Space Treaty requires “authorization and continuing supervision” by the launching state (UNOOSA, 1967, Art. VI), a principle historically predicated on real-time human control. When AI systems make decisions in milliseconds, this paradigm becomes operationally challenging. Technical assurance frameworks have begun to formalize evidence-backed operational boundaries, yet they remain legally non-binding (KRI, 2026; Benton, 2026). Conversely, legal proposals for algorithmic auditing lack technically feasible implementation tools (Bratu, Lodder & van der Linden, 2021; Bratu, 2024). This paper argues that a primary cause of the gap is the absence of an operational definition of Meaningful Human Control (MHC) in space law. The paper proposes a new understanding of MHC as an evidentiary standard—the verifiable capacity of a state to demonstrate that an autonomous system operated within pre-authorized boundaries—and proposes a hybrid legal-technical mechanism to operationalize it: ex-ante certification of AI decision boundaries combined with cryptographically signed runtime receipts. By embedding this mechanism into national launch licensing and commercial insurance requirements, the framework transforms technical evidence into a strong evidentiary basis for a rebuttable inference of compliance under Article VI within the proposed domestic framework. The mechanism is designed to improve political feasibility and provide a technically grounded response to the liability gap while preserving the primacy of the 1972 Liability Convention. To facilitate practical implementation, the paper provides three operational annexes: a template for Authority Profiles (Annex A), minimum data fields for cryptographic receipts (Annex B), and a glossary of key interdisciplinary terms (Annex C).