Māori AI is artificial intelligence that is created, governed, controlled and used in ways that uphold tino rangatiratanga, tikanga Māori and mātauranga Māori, and that protects Māori people, data and knowledge from harm.
This page brings together Dr Karaitiana Taiuru’s work on Māori AI sovereignty, the He Tangata, He Karetao, He Ātārangi framework, Māori AI ethics and algorithmic bias in Aotearoa New Zealand.
AI systems are already trained on te reo Māori, mātauranga Māori and the images, voices and stories of Māori, mostly without free, prior and informed consent. They are also being used to make decisions about Māori in health, justice, welfare, employment and education. Māori AI asks two questions of every system: who controls it, and who carries the risk?
What is Māori AI?
Māori AI has four connected parts:
- Māori AI sovereignty: Māori authority over the data, infrastructure, models, workforce and economic value of AI.
- Te Tiriti AI ethical principles: ethical principles for AI that are grounded in Te Tiriti o Waitangi rather than imported Western ethics alone.
- Māori AI ethics: applying tikanga Māori to consent, accountability and the use of mātauranga Māori in AI.
- Algorithmic bias: identifying and preventing AI systems that produce worse outcomes for Māori.
Māori AI sovereignty
Māori AI sovereignty is the ability of Māori to create and control AI under tino rangatiratanga, rather than being only customers of AI systems owned and governed offshore. It builds on Māori Data Sovereignty and extends it to the code, infrastructure, people and investment that AI depends on.
Tino rangatiratanga mō tātou, ā, mō ngā uri ā muri ake nei.
The ability to create and control our destiny for generations to come.
The Māori AI Sovereignty Principles, a localised version of the international Indigenous AI sovereignty principles, set out five pou:
- Data sovereignty and data governance: our data, our protocols. AI built on tikanga and mātauranga Māori with intergenerational consent.
- Control over infrastructure and code: if we don’t own it, we don’t control it. Code, servers and models operated under rangatiratanga.
- Skilled workforces: Māori AI and machine learning professionals able to build and deploy their own solutions.
- Economic reinvestment: redirecting digital value to strengthen the Māori economy and create well-paid Māori technology jobs.
- Investment in innovation: iwi, hapū, marae and Māori organisations investing in their own AI initiatives and research.
Read the Māori AI Sovereignty Principles →
New Zealand and Māori AI Sovereignty →
He Tangata, He Karetao, He Ātārangi: a kaupapa Māori AI framework
He Tangata, He Karetao, He Ātārangi (a person, a puppet, a shadow) is a kaupapa Māori framework developed by Dr Karaitiana Taiuru for understanding and describing the nature of an AI. Released in March 2026, it is one of the first Indigenous frameworks of its kind. Each part of the whakatauāki describes a different dimension of AI:
- He Tangata (a person): AI presents as a person. It speaks in natural language, responds to context and increasingly fills roles once held by human practitioners.
- He Karetao (a puppet): AI is moved by many hands at once: developers, operators, users and the unpredictable interactions between them. This can spread accountability so thinly that no one takes responsibility for harm.
- He Ātārangi (a shadow): AI is a shadow of human knowledge and culture. Mātauranga Māori in a training dataset keeps its tikanga and its mauri, but the AI itself cannot be held to account as a person or collective can.
The framework has three practical implications: consent from those whose knowledge trains an AI is a right, not a courtesy; accountability must extend across the full chain of developers, operators and deployers; and AI cannot be an authority on tikanga Māori or mātauranga Māori.
Read the He Tangata, He Karetao, He Ātārangi framework →
Te Tiriti AI ethical principles
Most AI principles used in New Zealand are adapted from international benchmarks such as the OECD AI Principles. Te Tiriti AI ethical principles start instead from Te Tiriti o Waitangi, treating it as a threshold every AI system must meet rather than a consultation step added at the end.
AI Principles in Aotearoa NZ →
Māori AI ethics
Māori AI ethics applies tikanga Māori to how AI is built and used. The central issues are consent for the use of te reo Māori and mātauranga Māori in training data, the protection of identity, moko and likeness, accountability when AI causes harm, and the risk that Māori knowledge is lost or appropriated through proprietary AI platforms.
- Record of stolen Māori knowledge by AI: an ongoing investigation into Māori books, music, video and research scraped into AI systems.
- Gen AI for ethical use of mātauranga: a template statement organisations can adopt (CC BY 4.0).
- Loss of mātauranga to AI: the risks of building reo and mātauranga models on paid, proprietary AI services.
- AI agents and legal personhood in New Zealand: a tikanga Māori analysis of agentic AI.
Algorithmic bias against Māori
Algorithmic bias occurs when an AI or automated system produces systematically worse outcomes for a group of people. For Māori, bias has been documented in facial recognition, health algorithms, recruitment tools, generative AI images and te reo Māori translation. Because Māori already face systemic inequities, biased AI compounds existing harm rather than creating new harm in isolation.
- Protecting Māori from AI and algorithmic bias
- Facial recognition, algorithmic bias and compounded harm to Māori and Pacific communities
- Bias in health AI
- AI bias with Māori in job interviews
- ChatGPT and its gender bias with te reo Māori
Māori AI FAQ
What is Māori AI sovereignty?
Māori AI sovereignty is the ability of Māori to create and control AI under tino rangatiratanga. It covers five pou: data sovereignty and data governance, control over infrastructure and code, skilled workforces, economic reinvestment, and investment in innovation.
What does He Tangata, He Karetao, He Ātārangi mean?
It means “a person, a puppet, a shadow”. It is a kaupapa Māori framework by Dr Karaitiana Taiuru that describes AI as presenting like a person, operating like a puppet moved by many hands, and being a shadow of human knowledge and culture.
Can AI be trained on te reo Māori and mātauranga Māori without consent?
Under Te Tiriti o Waitangi and the United Nations Declaration on the Rights of Indigenous Peoples, Māori have rights over their taonga, including their language and knowledge. Training AI on te reo Māori and mātauranga Māori without free, prior and informed consent breaches those rights, even where the material was publicly available online.
Is AI an authority on tikanga Māori?
No. AI is a shadow of the knowledge it was trained on and cannot be held to account. AI-generated content about tikanga Māori or mātauranga Māori must be treated with caution and must not replace the guidance of recognised knowledge holders.
How does algorithmic bias affect Māori?
Biased AI systems can misidentify Māori in facial recognition, under-serve Māori in health algorithms, disadvantage Māori job applicants and produce stereotyped images and translations. Because these systems are often trained on data that under-represents or misrepresents Māori, they can deepen existing inequities.
Is storing data in New Zealand enough for Māori AI sovereignty?
No. Storing data in New Zealand is data residency. Sovereignty also requires Māori authority over who accesses the data, how it is used, whether it trains AI models, and who owns the infrastructure, code and value that result.
How we help
Taiuru & Associates advises iwi, hapū, Māori organisations, government agencies, universities and technology companies on Māori AI sovereignty, AI ethics and bias, cultural risk in generative AI, and the protection of te reo Māori and mātauranga Māori in AI systems.
AI governance services →
Māori Data Sovereignty →
Original frameworks →
All Māori AI articles →
Last updated: 8 October 2026