Māori data sovereignty
The doctrine that Māori data is a taonga, the institutions behind it, and the one New Zealand example of sovereignty achieved at the data layer.
The kaupapa described here belongs to the organisations named.
The doctrine
Māori data sovereignty holds that Māori data is a taonga, subject to the rights guaranteed under te Tiriti o Waitangi, and that Māori data should be subject to Māori governance.
Te Mana Raraunga, the Māori Data Sovereignty Network, was established in 2015, and its charter was approved in April 2016. The charter states: “Data is a living tāonga and is of strategic value to Māori”. The network holds that “Māori data should be subject to Māori governance”. Its definition of Māori data is broad: data produced by Māori, and data about Māori and the environments they have relationships with.
Where this meets the ladder
In this module’s terms, the doctrine speaks first to layer five: the training data, meaning everything a model learned from. The big models have already absorbed te reo Māori at scale. Whisper, OpenAI’s speech-recognition model, was trained on 1,381 hours of spoken te reo Māori collected from the web, on the paper’s own figures. Te Hiku Media’s response, published on the Papa Reo blog on 24 January 2023, described Whisper as “another case study in colonisation” and objected that the recordings had been taken without consent.
The objection is a sovereignty claim in exactly this module’s sense. The community that created the recordings held none of the layers: their language had gone into a model they could no longer inspect, correct or govern.
The layer, held
Te Hiku Media, a Māori media organisation, built its own speech recognition for te reo Māori: software that turns spoken te reo into written text. The recordings it learned from were gathered with consent, through the Kōrero Māori campaign: more than 300 hours of speech from about 2,500 speakers, in roughly ten days. The models run on Papa Reo, Te Hiku’s language platform, backed by a $13 million MBIE award.
Access runs under the Kaitiakitanga License, Te Hiku’s own permission-based licence. Anyone wanting to use the code seeks permission, and Te Hiku decides, with priority given to Indigenous language revitalisation. Te Hiku describes the model as guardianship, in the sense of kaitiaki: guardian, protector, custodian.
Measured against this module’s ladder, Te Hiku holds the recordings, the models, the licence and the decision. As at September 2026 this is the clearest New Zealand example of the data layer held in practice.
The institutions
- The Mana Ōrite Relationship Agreement (signed 30 October 2019) between Stats NZ and the Data Iwi Leaders Group establishes an equal-status partnership in which each party’s knowledge carries “equal explanatory power” in decisions about the data system.
- The Māori AI Governance Framework, unveiled by Te Kāhui Raraunga on 30 October 2025, extends the 2023 Māori Data Governance Model to AI systems and urges its adoption across government.
- The Government’s Responsible AI Guidance for the Public Service (GenAI) carries a dedicated section on Māori, Pacific peoples, ethnic communities and generative AI.
- Te Pā Tūwatawata, a Māori-owned decentralised data-storage network built by Te Kāhui Raraunga, was reported by RNZ in May 2026 as a step towards holding the storage layer itself.
What this means for readers
If your organisation holds Māori data and is weighing AI tools, this doctrine bears on the decision. The Māori AI Governance Framework is the current reference point. Public servants have the dedicated GenAI guidance section. The questions on the New Zealand page apply here. One more belongs at the front: whose data is this, and under whose governance does it sit?
Nothing on this page is legal advice. Nothing on it is a statement of tikanga.
Sources. Each carries a live link and a dated Perma.cc record.
- Te Mana Raraunga charter, approved April 2016 — record
- “OpenAI’s Whisper is another case study in Colonisation”, Papa Reo blog, 24 January 2023 — record
- Kaitiakitanga License, Te Hiku Media — record
- Mana Ōrite Relationship Agreement, Stats NZ and the Data Iwi Leaders Group, 30 October 2019 — record
- Māori AI Governance Framework, Te Kāhui Raraunga, 30 October 2025 — record
- Te Pā Tūwatawata, Te Kāhui Raraunga — record
- RNZ, Māori-owned data storage network, May 2026 — record
All checked and captured 23 September 2026.