From Principles to Practice: Māori Data Sovereignty at the Point of AI System Design

This working paper defines an implementation problem at the intersection of Māori data sovereignty and AI system design. AI governance offers principles, standards, and lifecycle tools. Māori data sovereignty offers operational instruments, governance models, and working infrastructure. The unresolved problem is narrower than a gap in frameworks. Small teams now build AI products on commercial infrastructure they do not control, with capability, leverage, and continuity that differ sharply from the institutions most frameworks assume. The paper examines the conjunction of four conditions at the point of system design: recognised Māori authority, unequal organisational capability, inherited commercial infrastructure, and organisational change including investment, acquisition, and insolvency. It introduces the concept of governance debt, sets out five design-time tensions any credible approach must resolve, and presents six normative propositions and a research programme. The 2025 bankruptcy and sale of 23andMe illustrates the structural risk of deferring governance until after data has moved. Native Sentient is developing a practical response, the Governance Translation Layer. This paper names that work and defines the problem it must solve. The operational architecture is reserved for Māori-governed testing. This is a working paper released for discussion and feedback. It has not been peer reviewed. It is intended for founders, product teams, investors, policymakers, and researchers working on Indigenous data sovereignty and AI governance.