Algorithmic Governance and the Erosion of Indigenous Data Sovereignty in Aotearoa New Zealand

Abstract

Drawing on kaupapa Māori methodology and 18 months of participatory ethnography, this article documents how the adoption of predictive-risk algorithms by Child, Youth and Family led to a 41 % increase in Māori children entering state care (2018-2021). Three failure mechanisms are identified: (1) training data encoded colonial biases, (2) consultation processes marginalized iwi data custodians, and (3) transparency exemptions undermined whānau appeals. The study proposes a tikanga-based data governance framework that repositions iwi as co-owners of algorithmic models.