Over the last few months, Aotearoa has started moving AI out of the demo phase and into ordinary systems. Wellington has a Public Service AI Framework, but it isn’t binding, and the wider regulatory posture is still deliberately light-touch. The government is now funding an AI Advisory Pilot to push uptake in business, while Health New Zealand has already endorsed an AI-powered wellbeing guide. At the same time, Te Kāhui Raraunga has been warning that agencies still need bias monitoring, algorithm transparency, public registers, and a way to shut systems down when they start reinforcing prejudice. In Waikato, researchers are pushing a different path, keeping Māori language technology inside Māori-led environments where authority over data and dialect doesn’t disappear into someone else’s stack. That’s the real pressure point. AI adoption is moving. Authority is still patchy.
On the surface we have alot of problem solving theory in place. Most AI teams can show a workflow, a contract, and a privacy statement. That’s the easy part. The harder question is often the one we avoid through colloquialisms, couching statements, and ambiguity: who had authority to approve the use of Māori data, what uses were actually permitted, what context had to travel with that material, and what limits still apply after the data has been cleaned, linked, labelled, embedded, queried, or pushed into downstream systems. Māori data sovereignty has never treated that as an optional layer. It treats Māori data as subject to Māori governance, and it extends control across creation, collection, access, analysis, interpretation, management, security, dissemination, use, and reuse (Te Mana Raraunga, 2018).
Extraction, in this setting, isn’t limited to scraping or outright theft. It happens when Indigenous data, language, stories, knowledge, or social relations are turned into institutional value while decision-making power over collection, classification, access, interpretation, and reuse sits somewhere else. That is the problem Indigenous data sovereignty was built to confront. The field emerged in response to poor data practices running from the design of data items through to reporting, and it treats data as a strategic resource tied to self-determination, governance, and the ability to set terms before value is taken rather than after harm has already landed (Kukutai & Taylor, 2016; Lovett et al., 2019; Kukutai, 2020).
Those rights don’t shrink once the material becomes digital. UNDRIP protects Indigenous participation in decisions affecting their rights, requires good-faith consultation toward free, prior, and informed consent for measures that may affect them, and recognises the right to maintain, control, protect, and develop cultural heritage and traditional knowledge (United Nations, 2007). Te Mana Raraunga carries that into Aotearoa’s data context by requiring provenance, purpose, collection context, and the parties involved to remain visible in metadata, and by warning that present decisions about data can produce long-tail effects for future generations (Te Mana Raraunga, 2018). Once those obligations are flattened into a generic permission model, “reuse” starts looking clean on paper while control has already slipped.
That slippage gets worse in state and research systems because datafication changes how people are seen and sorted. Māori data sovereignty scholarship in Aotearoa has been clear that state categories, research conventions, and data infrastructures don’t passively describe Māori realities. They shape them. That’s part of why data sovereignty can’t be reduced to storage location or database access. It also has to confront classification, representation, linkage, surveillance, and the power to decide what counts as legitimate knowledge in the first place (Kukutai & Cormack, 2019; Cormack & Kukutai, 2022). Māori data governance work from a Māori world view places that problem inside a longer history of research harm and weak Indigenous control, a pressure that has only intensified as digital systems accelerate the pace of collection and reuse (West, Hudson, & Kukutai, 2020).
AI doesn’t create an exemption from any of that. It increases the number of places where control can disappear. Māori algorithmic sovereignty work makes the bridge explicit by treating algorithms as a use of data, which means sovereignty questions follow the data into ranking, prediction, optimization, recommendation, and automated decision-making rather than stopping at collection (Brown et al., 2024). Work on Māori trust in automated decision-making pushes in the same direction by treating data, models, and system governance as connected rather than separable, and by insisting that Māori participation in design, governance, and accountability can’t be bolted on after deployment (West et al., 2020). If a model has been trained, tuned, or operationalized on Māori data without clear authority, context, and governance, the pipeline hasn’t somehow become neutral just because the output looks technical.
The reason this gets overlooked so often is boring, structural, and common. Reuse culture still dominates. CARE was developed because mainstream data frameworks keep privileging discovery, interoperability, and reuse while Indigenous governance requires collective benefit, authority to control, responsibility, and attention to future use (Carroll et al., 2020). More recent work on aligning CARE with FAIR warns that institutions can reference Indigenous governance principles without changing their own protocols, policies, or power structures, which leaves Indigenous authority weak while the language looks progressive (Taitingfong et al., 2024). On top of that, privacy paperwork still gets used as a substitute for governance even though Māori data sovereignty and privacy work has already argued that personal data protection is necessary but insufficient because current privacy regimes don’t recognise or protect collective Indigenous privacy rights (Kukutai et al., 2023). The practice gap is visible in the wider literature too. A 2024 scoping review of Indigenous health-data research found that ethics approval was reported in 93% of studies, while Indigenous guiding principles appeared in 21%, data sovereignty in 41%, and consent in 33% (Engstrom et al., 2024).
Aotearoa already has enough evidence that this isn’t hypothetical. Work on Māori and the Integrated Data Infrastructure shows that even de-identified, whole-population administrative systems still raise live questions for Māori about governance, aspiration, and control, and it sets out steps needed to realise Māori data aspirations inside that environment (Greaves et al., 2024). Research on Māori data sovereignty in public and private sector organizations lands in a similar place, highlighting whanaungatanga, rangatiratanga, kotahitanga, and akoranga as practical principles for organizational data culture rather than abstract values work (Lilley et al., 2024). Health governance scholarship has pushed the point even further by building an explicit planning and protocol checklist because the broader Māori Data Governance Model still hasn’t been meaningfully implemented in many settings, especially across the full data-handling lifecycle (Kremer et al., 2025). The pattern is obvious enough. Organizations know the language faster than they change control.
That’s why extraction matters so much here. Once Māori data enters a pipeline, value can keep moving downstream even while authority fades upstream. The model improves, the service gets cheaper, the dashboard gets smarter, the vendor pitch gets cleaner, and the system still can’t answer basic questions about mandate, provenance, participation, conditions of use, and limits on reuse. When that happens, the branding has improved, the control story hasn’t. Māori data sovereignty asks a harder standard and a better one: who decides, whose interests are carried, who benefits, what remains visible, and what stays off limits. If an AI pipeline can’t answer those questions clearly, it hasn’t solved the governance problem. It’s just made extraction easier to package (Te Mana Raraunga, 2018; Carroll et al., 2020; Brown et al., 2024; Greaves et al., 2024; Lilley et al., 2024).
References
Brown, P. T., Wilson, D., West, K., Escott, K.-R., Basabas, K., Ritchie, B., Lucas, D., Taia, I., Kusabs, N., & Keegan, T. T. (2024). Māori algorithmic sovereignty: Idea, principles, and use. Data Science Journal, 23, Article 15. doi:10.5334/dsj-2024-015
Carroll, S. R., Garba, I., Figueroa-Rodriguez, O. L., Holbrook, J., Lovett, R., Materechera, S., Parsons, M., Raseroka, K., Rodriguez-Lonebear, D., Rowe, R., Sara, R., Walker, J. D., Anderson, J., & Hudson, M. (2020). The CARE principles for Indigenous data governance. Data Science Journal, 19(43), 1-12. doi:10.5334/dsj-2020-043
Cormack, D., & Kukutai, T. (2022). Indigenous peoples, data, and the coloniality of surveillance. In A. Hepp, J. Jarke, & L. Kramp (Eds.), New perspectives in critical data studies: The ambivalences of data power (pp. 121-141). Springer. doi:10.1007/978-3-030-96180-0_6
Engstrom, T., Lobo, E. H., Watego, K., Nelson, C., Wang, J., Wong, H., Kim, S. L., Oh, S. I., Lawley, M., Gorse, A.-D., Ward, J., & Sullivan, C. (2024). Indigenous data governance approaches applied in research using routinely collected health data: A scoping review. npj Digital Medicine, 7, Article 68. doi:10.1038/s41746-024-01070-3
Greaves, L. M., Barlow, F. K., Kukutai, T., Ruwhiu, D., Hayward, J., Pega, F., Barnes, H. M., Borell, B., Campbell, S., Sheridan, J., Moewaka Barnes, A., McCreanor, T., & Sibley, C. G. (2024). Māori and the Integrated Data Infrastructure: An assessment of the data system and suggestions to realise Māori data aspirations. Journal of the Royal Society of New Zealand, 54(1), 1-17. doi:10.1080/03036758.2022.2154368
Kremer, L., Dewes, O., Dineen, R., Ruru, R., Murchie, M., Manuela, S., Elers, P., Wilson, D., Sullivan, P., Pomare, P., & Huria, L. T. (2025). The Māori health data governance model: A planning and protocol checklist. Health Promotion Journal of Australia. doi:10.1002/hpja.70018
Kukutai, T. (2020). Indigenous data sovereignty. In D. Vinding & C. Mikkelsen (Eds.), The Indigenous World 2020 (pp. 655-659). International Work Group for Indigenous Affairs.
Kukutai, T., & Cormack, D. (2019). Mana motuhake ā-raraunga: Datafication and social science research in Aotearoa. Kotuitui: New Zealand Journal of Social Sciences Online, 14(2), 131-139. doi:10.1080/1177083X.2019.1648304
Kukutai, T., & Taylor, J. (2016). Data sovereignty for Indigenous peoples: Current practice and future needs. In T. Kukutai & J. Taylor (Eds.), Indigenous data sovereignty: Toward an agenda (pp. 1-22). ANU Press. doi:10.22459/CAEPR38.11.2016.01
Kukutai, T., Teague, V., Tuffery Huria, L., Watts, D., & Sterling, R. (2023). Māori data sovereignty and privacy. Te Ngira Institute for Population Research.
Lilley, S., Cunningham, C., Kukutai, T., Mika, C., Finau, G., & Parekura, S. (2024). Māori data sovereignty: Contributions to data cultures in the government sector in New Zealand. Information, Communication & Society. doi:10.1080/1369118X.2024.2302987
Lovett, R., Lee, V., Kukutai, T., Cormack, D., Carroll Rainie, S., & Walker, J. D. (2019). Good data practices for Indigenous data sovereignty and governance. In A. Daly, S. K. Devitt, & M. Mann (Eds.), Good data (pp. 26-36). Institute of Network Cultures.
Taitingfong, R., Martinez, A., Hudson, M., Lovett, R., Maher, B., Prehn, J., Rowe, R. K., Boileau, K., Franks, A., Khan, S., Walker, J. D., & Carroll, S. R. (2024). Aligning policy and practice to implement CARE with FAIR through Indigenous Peoples’ protocols. Acta Borealia, 41(2), 80-90. doi:10.1080/08003831.2024.2410112
Te Mana Raraunga. (2018). Principles of Māori data sovereignty. Te Mana Raraunga.
United Nations. (2007). United Nations Declaration on the Rights of Indigenous Peoples. United Nations.
West, K., Hudson, M., & Kukutai, T. (2020). Data ethics and data governance from a Māori world view. In K. Williamson & K. DeSouza (Eds.), Reimagining data governance: A data-centric framework for ethical and effective use of data (pp. 53-77). Emerald Publishing.
West, K., Wilson, D., Thompson, A., & Hudson, M. (2020). Māori perspectives on trust and automated decision-making. Report for the Digital Council Aotearoa New Zealand.