--- title: Govt AI Framework description: Explore the importance of Māori data consultation in AI governance and policy for effective regulatory practices. url: https://www.taiuru.co.nz/govt-ai-framewor/ markdown_url: https://www.taiuru.co.nz/govt-ai-framewor.md type: post slug: govt-ai-framewor author: Karaitiana Taiuru published: 2026-09-15T03:25:27Z modified: 2026-10-23T06:22:53Z language: en-NZ taxonomies: category: - AI Governance word_count: 2501 tokens: 4132 --- # Govt AI Framework CASE STUDY –The need for Māori Expertise in AI Governance and Policy and Frameworks. In May 2026 the Ministry for Regulation published guidance on the use of Artificial Intelligence (AI) in regulatory practice Practical steps for regulators leading AI innovation with confidence. This case study is based on Version 1 of that document. The guidance lightly touched over Te Tiriti o Waitangi, Māori Data, and Māori Data Sovereignty. To strengthen the document before release, the Ministry commissioned an independent expert peer review by a non-Māori company with no expertise in te ao Māori. At no point did the Ministry engage a Māori reviewer or Māori data governance expertise. Nor did the independent expert peer review company engage with Māori expertise. That review was carried out by an AI consultancy which was scoped for a non-technical audience. The one substantive recommendation the Ministry did receive was to bring Māori expertise in, was made by the AI reviewer but was set aside as ‘outside our skillset’ and a risk to the publication timeline. Nevertheless, the review and publication proceeded without Māori expertise because doing so would affect scope and timing. In New Zealand, the issues with employers relying on Māori staff for Māori representation is well documented and government agencies relying on Māori groups without authority and relevant skills, while not new, is increasingly growing with the introduction of Artificial Intelligence systems. The AI Forum statistics show a reasonable increase with organisations seeking Māori expertise and input to AI projects, though it is not clear if this is genuine consultation. REF The intention of this case study is to highlight the need for organisations and in particular the government to consult with Māori and indeed other stakeholder impacted communities when considering and deploying AI systems. There are volumes of international examples of AI development not considering and involving stakeholders and the social and economic impacts including brand reputation, risks to communities of bias, exclusion, harm, human rights impacts and system failures.   Disclaimer This case study is drawn entirely (unless stated otherwise) from documents released by the Ministry for Regulation (Te Manatū Waeture) under the Official Information Act 1982, comprising the request itself (Appendix A (pdf)), the LensenMcGavin AI peer-review engagement and report (Appendix B (pdf), and the internal feedback and decision emails (Appendix C (pdf). It is written to illustrate a recurring failure pattern in public-sector AI policy, not to single out individuals; the names appear only because the documents do.   Analysis a) The purpose of the review The earliest document (Appendix C, 5 February 2026) is internal feedback from an official to a colleague. It is candid about why external review was wanted at all: The point of review is to be able to say a review happened. “…it helps to say this has been external reviewed by XX. This is good practice which is why we have done something similar with each of our sector review reports…” Appendix C, internal email, 5 February 2026 b)  Māori data sovereignty The same email records the Ministry’s working posture on the Māori content: “I do not believe Māori data sovereignty is well understood across the public service yet and we do not want to try and define it in the document. However, I suggest we add some links to the authoritative source… as the crown have defined it.” Appendix C, internal email, 5 February 2026 First, poor understanding across the public service is treated as a reason to avoid the subject rather than a reason to seek those who do understand it. Second, the authority to be cited is ‘the Crown’ the Treaty partner whose exercise of power is what Māori Data Sovereignty exists to constrain. Finally, the Crown has no one united definition, despite a clear definition by the Waitangi Tribunal. Deferring the definition of Māori Data Sovereignty to the Crown’s own framing inverts the concept. Tino rangatiratanga over Māori data does not derive from a Crown definition of it. c) The independent expert The reviewer engaged LensenMcGavin AI (Appendix B): LensenMcGavin AI describes itself as an AI firm, and its own critique of the draft was that passages: “come across as very ‘ChatGPT’”. The report’s own disclosure statement records: “Generative AI tools (e.g. Gemini) were used to refine writing and structure. … All suggestions, critiques, and comments were produced by the authors. All GenAI output was reviewed and edited by the authors.” The subsequent courtesy review reads: “No notes from our end. Absolutely brilliant… without wanting to sound too much like a sycophantic chatbot myself… you will have produced the gold standard for public sector AI guidance.” The subsequent correspondence is strongly positive about the revised document. Whatever view is taken of the reviewer’s independence, the more significant issue is that the engagement did not supply the Māori-specific expertise subsequently identified as necessary. The contract (Appendix B) scoped the deliverable as feedback: “Suitable for a senior regulatory audience with a non-technical or IT background or capability” The engagement was, by its own terms, a light-touch readability and framing pass, not a specialist review of the document’s most consequential dimensions. d) The AI reviewer advised to engage Māori experts The point in the record is that the inadequacy of the Māori content was flagged by the AI reviewer itself. Section 4.6 of the LensenMcGavin report (Appendix B) states: “There is a real risk that in their current state, the concepts will function merely as lip-service and sign posting… Māori will disproportionately feel the negative impacts of AI… we strongly recommend that MfR engages with experts and leaders in this domain to properly inform these components of their guidance.” Appendix B, LensenMcGavin AI peer review, s 4.6 The recommendation came from a non-Māori AI consultancy which is itself the point. The Ministry was willing to pay for and act on an external voice; it simply chose the wrong external voice for this content, and then, when that voice pointed to the right one, declined to follow it. e) Recommendation to engage Māori was treated as cost, not obligation Appendix C records how the Ministry resolved the question. The lead official’s own framing of the choice (12 April 2026) was: “1. Keep this guidance high-level and refer to system guidance (preferred and as currently drafted), or 2. Expand this section further… recognising that further expansion is outside our skillset and would require consultation by GCDO and TPK adding delay to publication of this guidance.” Appendix C, internal email, 12 April 2026 Engaging properly is described as: “outside our skillset” and as “adding delay.” ‘Outside our skillset’ is why external Māori expertise was needed, in the same way that being non-technical was treated as a reason to buy in AI expertise. The Ministry applied the buy-in logic to the technical gap and the avoidance logic to the Te Tiriti gap. The settled outcome (14 April) was a Te Tiriti section written without going deep into Māori data sovereignty at this stage, proposed to be tested with Te Puni Kōkiri (TPK) only, consultation on the Crown’s own drafting, after the fact, rather than Māori partnership in shaping it.   The wrong way to engage a) Te Tiriti as a threshold The Ministry’s decision treated the Māori content as one refinement among many, weighed against timeline, skillset, and scope like a grammar fix or a missing case study. It is a constitutional relationship that conditions how the Crown may design and deploy decision-making systems, not an optional enhancement to a document about those systems. b) The architecture A simple test that separates genuine engagement from its appearance include: are the Māori provisions structurally operative do they change what a regulator must do, or are they tokenistic, so the document can be seen to mention them? The AI reviewer, without using these words suggested: “lip-service and sign posting.” The Ministry’s chosen path to keep it high-level, link to Crown sources, avoid definition and defer depth is the path that produces tokenism. Tokenism can only be fixed by letting Māori expertise in be a part of the life cycle and review.   c) Tino rangatiratanga Proper engagement locates authority with those who hold it: hapū, iwi and Māori data governance experts, brought in at the conception stage of the life cycle to influence scope, not consulted late enough only to react. d) Data residency is not data sovereignty The guidance and its review circle around where data is stored and sovereignty where relevant. These conflate two different things. Data residency is a question of servers and jurisdiction is where the data physically sit. Māori data sovereignty is a question of authority, who decides how Māori data is collected, used, and governed, and on whose terms. An AI reviewer scoped for a non-technical audience is not positioned to hold that distinction, and the record shows it did not. Only Māori expertise reliably does this while procuring the wrong reviewer produced the wrong frame. e) The disproportionate harm The review states that ‘Māori will disproportionately feel the negative impacts of AI.’ The resulting process created an inversion where the group facing the most harm received the least engagement, because engaging them was slower. The process appears to have prioritised publication speed despite the reviewer’s warning about disproportionate impacts on Māori.   3.6 A Māori AI governance test A proposed operational governance framework could include: Is Māori content merely present? Has Māori expertise informed it? Can Māori expertise change the substance? Does the guidance require regulators to do anything differently? Who determines whether those requirements have been met? What proper engagement could have looked like Point by point, the record shows what was done and what partnership would have required instead. What the Ministry did What proper engagement requires Commissioned a general AI consultancy for the whole document, including the Māori content. Commission a separate, parallel review of the Te Tiriti and Māori Data sections by a reviewer with Māori Data Governance expertise. Sought external review to be able to say review had occurred. Seek review to change the document, and choose reviewers whose mana makes the content correct, not merely defensible. Treated outside our skillset as a reason to stay high-level. Treat outside our skillset as the trigger to bring in those whose skillset it is. The same logic used to justify buying AI expertise. Deferred Māori Data Sovereignty depth to avoid publication delay. Sequence the work so Māori engagement shapes scope from the start; publication timing follows the obligation. Planned to test finished drafting with Te Puni Kōkiri (a Crown agency). Engage Māori data governance expertise at the inception of the policy development process, with Te Puni Kōkiri and other Crown Māori capability providing appropriate institutional support where relevant. Let the draft use taonga and sovereignty as vocabulary. Let Māori expertise make those words structurally operative, changing what a regulator must do, not how the guidance reads. Kaupapa Māori researchers and Māori leaders continually raise the issues that Maori guidance is increasingly being decoupled from traditional Māori beliefs, even by some Māori groups. This publication goes to the extreme and ignored Māori guidance all together. No Māori organisation, iwi, hapū, rōpū, Māori Data Governance body, or individual Māori expert at all were consulted. Relevance to AI governance AI systems can reproduce and amplify existing inequalities. AI policy and guidance have historically neglected to properly integrate Māori principles, despite the potential for Māori to disproportionately experience negative impacts from AI. The consequences can arise at several levels. Data Māori Data and bias data may be incorporated into datasets without adequate consideration of Māori authority, provenance or governance. Models Models may encode historical patterns or biases that negatively affect Māori. Procurement Government agencies may procure systems without sufficient visibility of training data, testing data, model development or vendor practices. Decision-making Human-in-the-loop (HITL) arrangements may provide an appearance of human control without ensuring that decision-makers have sufficient technical, cultural or domain expertise. The reviewer explicitly warned that HITL is not enough without review and audit processes capable of addressing human error and automation bias. Accountability The guidance itself states that accountability remains with organisations and legally authorised decision-makers. The reviewer considered this point sufficiently important that it should be made much more prominent. A human decision-maker cannot provide meaningful oversight of an AI system if they do not understand the system’s limitations, the provenance of its data, the risks associated with its outputs, and the cultural and legal context in which those outputs are being applied.   Lessons for government AI procurement The case also raises a broader procurement lesson. The Ministry’s contract was carefully scoped around the needs of senior regulatory leaders, regulatory defensibility, clarity, practical applicability and technical accuracy. But the subject matter itself required additional expertise. This suggests that government AI procurement should ask, at the outset: What expertise is required to properly evaluate this AI system or policy? That may include: AI technical expertise information security privacy legal expertise domain expertise Māori data governance expertise Te Tiriti expertise accessibility expertise affected-community knowledge. The appropriate expertise should be determined by the subject matter and risk profile, not simply by the procurement category. Otherwise, an agency can technically satisfy a requirement for “independent peer review” while still failing to obtain the specialist expertise required to properly scrutinise the most consequential parts of the work.   Conclusion The Ministry for Regulation’s AI guidance case provides an unusually clear example of the difference between acknowledging Māori interests and embedding Māori expertise. The Ministry commissioned an independent external review of its AI guidance. That review identified numerous technical, governance and regulatory issues. Among them was a specific and substantive concern about Māori Data, Māori Data Sovereignty and Te Tiriti o Waitangi. The reviewer didn’t merely suggest adding a paragraph. It warned that the existing treatment risked becoming “lip-service and sign posting” and strongly recommended engaging Māori expertise. The Ministry subsequently included Te Tiriti considerations in the near-final guidance but indicated that the material would be expanded following future sector feedback and support from Te Puni Kōkiri. If Māori Data Sovereignty and Te Tiriti were sufficiently important to appear in government AI guidance, and an independent reviewer specifically identified the need for Māori experts and stakeholders, why was that specialist expertise not demonstrably embedded in the development and review process from the outset? Māori expertise should not be an amendment to AI governance. It should be part of its life cycle.   Post publication and non OIA related. The author of this case study criticised the publication and the independent reviewer on LinkedIn. There was a mixed reaction, though largely in support of the critiques. The Ministry contacted the author the next day and a meeting was held to discuss issues. The Ministry were open to feedback and the author provided written feedback that was included in the second version of the publication which is at the time of writing the latest version. However, this case study and critiques are still relevant, and the publication still lacks proper Māori perspectives and Te Tiriti acknowledgement.