Does the Responsible AI Guidance Adequately Protect Māori in Predictive Workplace Safety Regulation?

Abstract

Artificial intelligence (AI) is increasingly embedded in the design and delivery of public services in Aotearoa New Zealand, particularly in high-risk regulatory environments such as workplace health and safety. This paper evaluates the sufficiency of the Responsible AI Guidance for the Public Service in upholding Te Tiriti o Waitangi, applying three intersecting frameworks: Critical Tiriti Analysis, really responsive regulation theory, and post-colonial legal analysis. Drawing on the Woolworths warehouse strike in Australia and WorkSafe New Zealand's engagement with AI-enabled tools through enforceable undertakings, the paper critiques the Guidance's reliance on non-binding, soft-law principles. Across all five Critical Tiriti Analysis criteria, the Guidance scores either Silent or Poor. It lacks enforceable protections, Māori governance structures, and procedural fairness mechanisms. The paper proposes a series of reforms to embed co-governance, Māori Data Sovereignty, and legally mandated safeguards into public sector AI infrastructure.