State of AI in New Zealand
Updated: 10 June 2026
Executive Summary
The past month in Aotearoa New Zealand was not defined by a single blockbuster AI announcement. Instead, it looked like a practical adoption month: government expanded hands-on support for small firms using AI, health moved breast-screening AI from exploration into procurement and validation, Digital.govt.nz surfaced more concrete public-service AI tooling, and the National Cyber Security Centre treated “frontier AI” as an immediate cyber-readiness issue for government agencies. At the same time, trust remained the system-wide constraint: the AI Forum refreshed its national blueprint around social licence and sustainable AI, while One NZ’s new trust research found mainstream usage but a more skeptical public. (mbie.govt.nz)
That makes May to early June 2026 feel broader but not deeper than the March snapshot. More sectors are now doing real AI work, but mostly in bounded, operational ways: business mentoring, telehealth triage support, workflow tools, recordkeeping, and clinical-assist procurement. What is still missing is a public resolution of the national AI Research Platform contest, a fresh compute-infrastructure milestone on the scale of Datagrid’s March push, or a wave of major local startup funding that would change the national picture overnight. (mbie.govt.nz)
What Happened in the Past Month
Government shifted from AI strategy to practical adoption support
The clearest new government move was MBIE’s 15 May expansion of the AI Advisory Pilot for small businesses. The programme, delivered through the Regional Business Partner Network, was expanded from 50 to 150 businesses, extended to 31 January 2027, and offers eligible firms up to 50% co-funding capped at $15,000 to identify and implement practical AI use cases. The announcement came at the launch of the Great New Zealand AI Roadshow, a nationwide series aimed at moving firms from curiosity to implementation. (mbie.govt.nz)
In the public sector, the noteworthy signal was not a flashy new deployment but more operational scaffolding. By 9 June, Digital.govt.nz had surfaced a consolidated Public Service AI Toolkit that brings together agency resources including an AI policy template and records-management guidance. The records guidance is especially telling: it stresses that AI does not change what counts as a public record, and that prompts, drafts, summaries, chatbot interactions, and pilot outputs may all need to be retained where they materially affect decisions or service delivery. That is a sign the public service is moving from “can we use AI?” to “how do we run it properly?” (digital.govt.nz)
A smaller but meaningful regional signal also appeared in the Australia-New Zealand Leaders’ Meeting joint statement on 7 June, which said both countries see value in a close trans-Tasman relationship on AI to embed targeted safety and security measures, support adoption across the economy, and promote responsible AI practices globally. It is not a programme on its own, but it does show AI now appearing in top-level bilateral economic and security framing. (beehive.govt.nz)
Health remained New Zealand’s most visible AI deployment sector
Health kept moving, but the emphasis shifted from scribes to broader clinical and service workflows. On 27 May, the Government said procurement is now underway to select a preferred AI mammogram reading tool for testing and validation within BreastScreen Aotearoa, with a planned rollout from early 2027. Ministers said the system would perform one of the two independent reads in the mammogram process, while clinicians remain central to diagnosis and follow-up decisions. That is a material step beyond February’s exploratory call: breast-screening AI is now moving into procurement and validation. (beehive.govt.nz)
Outside hospital settings, Whakarongorau Aotearoa / New Zealand Telehealth Services rolled out an Azure AI-powered Welcome service in May for the 1737 text service and Women’s Refuge webchat. The system is designed to gather context, clearly disclose that users are interacting with AI, offer non-clinical support while people wait, and hand over to human staff with better background information. It is a good example of the month’s dominant pattern: AI being inserted into narrow workflow gaps to reduce waiting friction and administrative load, rather than replacing core human judgement. (news.microsoft.com)
Governance, cyber readiness, and public trust moved further toward the centre
The strongest official governance signal came from the National Cyber Security Centre on 4 June, when it published whole-of-government guidance on “cyber readiness in the Frontier AI era.” The advisory says frontier models create a “dual-use challenge” and a potential “vulnerability storm,” and tells government entities to prioritise existing cyber fundamentals, executive accountability, and remediation of gaps that could be exploited by AI-enabled threat actors. That is an important shift in tone: advanced AI is now being treated as an immediate security-operating issue, not just a future policy topic. (ncsc.govt.nz)
Another governance-adjacent development came from welfare administration. Parliament passed the Social Security (Modernisation) Amendment Bill on 29 May, widening the use of automated decision-making for straightforward welfare processes while adding requirements around bias management, transparency, human oversight, and challenge rights. Ministers explicitly noted that this is not generative AI, but it is still relevant to New Zealand’s AI story because it shows algorithmic decisioning continuing to move into core public administration under a more formal safeguards regime. MSD’s updated 2026 Automated Decision-Making Standard similarly emphasises transparency, public disclosure, human appeal channels, and bias controls. (beehive.govt.nz)
Trust also stayed front and centre outside government. One NZ’s AI Trust Report 2026, released 26 May, found 76% of New Zealanders had used an AI-powered tool or service in the past year, but fewer than two in five believed AI would deliver better outcomes for society; 62% said they would stop using a product or service if they were concerned about an organisation’s AI use, and 68% said human access would make them more comfortable with AI-powered customer interactions. One NZ responded by launching an AI Trust Hub. Whatever the exact methodology, the direction is consistent with other recent NZ indicators: usage is mainstreaming faster than trust. (media.one.nz)
Ecosystem activity was real, but it leaned toward capability and framing
The main ecosystem document this month was the AI Forum’s AI Blueprint for Aotearoa, refreshed on 6 May. It adds stronger emphasis on social licence and sustainable AI, and explicitly describes New Zealand as a “high-use, low-trust” environment. The associated work programme points to discussion series on sustainable AI, continued productivity research, and national convening activity rather than a single new institution or fund. That matters because it shows the ecosystem’s centre of gravity continuing to move toward trust, sovereignty, capability, and infrastructure conditions rather than pure hype about tools. (aiforum.org.nz)
Universities were active too, but mostly in applied and capability-building ways. The University of Canterbury announced an AI-powered soil-moisture system for farmers on 13 May, and on 2 June highlighted its Scaffolded AI Literacy (SAIL) framework, which aims to define AI literacy for Aotearoa in a way that includes Māori and Pacific perspectives. These are useful signs of applied research and education activity, but they do not yet amount to a national-scale research-commercialisation breakthrough. (canterbury.ac.nz)
Enterprise signals existed, but they were selective
On the private-sector side, the clearest public market signal came from Wellington-founded Blackpearl Group, which reported record FY26 ARR of $26.8 million, up 114% year on year, and positioned its Pearl Engine as an AI-driven commercial data platform. That is a meaningful growth story for a listed New Zealand software company, but it still looks like an outlier rather than evidence of a broad wave of AI-native commercial breakout across the NZ market. (nzx.com)
Trend Line Across Recent Snapshots
Compared with the 2 April 2026 snapshot, the texture has shifted again. March was mainly about infrastructure, research-platform mechanics, privacy guidance, and enterprise risk. The current period is more about applied adoption support, bounded public-service tooling, health workflow expansion, and cyber-readiness. In other words, the story has moved from “building the conditions” to “using those conditions in more ordinary places,” even if only in carefully constrained ways. (mbie.govt.nz)
The broader pattern across recent snapshots is that New Zealand’s AI activity is broadening horizontally. Health is still the most visible deployment sector. Government is adding practical templates and support mechanisms. Small businesses are getting more structured help. Trust, sovereignty, and safety remain recurring themes. What New Zealand still lacks is the same level of momentum in large-scale research-commercialisation decisions, venture funding, and locally anchored frontier infrastructure execution. (beehive.govt.nz)
What Looks Quiet, Unchanged, or Early
The biggest “missing” development is the national AI Research Platform outcome. MBIE’s platform page still says only that phase-two proposals were due on 31 March 2026, assessed in mid-April, and that “an update on a timeline of announcements will be made in due course.” As of 10 June, that means the public winner or final funding shape still had not visibly emerged on the main official page reviewed for this snapshot. For a programme backed by up to $70 million over seven years, that absence matters. (mbie.govt.nz)
Infrastructure was also quieter than in March. In the sources reviewed for this update, there was no new Datagrid-scale compute or data-centre milestone comparable to the previous snapshot’s Southland consent-and-power story. The infrastructure conversation did continue indirectly through the AI Forum’s sustainable-AI framing and through data-sovereignty discussion, but the month’s most concrete action was elsewhere. (aiforum.org.nz)
Commercially, the evidence is still mixed. Blackpearl stood out, and there were practical vendor and service announcements, but there was no obvious cluster of major New Zealand AI fundraising rounds, large new product launches, or national enterprise rollouts that would suggest a step-change in commercial scale. Education activity, likewise, remains strongest in AI literacy, governance, and executive capability-building rather than in system-wide institutional transformation. (nzx.com)
Overall Assessment
The simplest reading of the past month is that AI in Aotearoa New Zealand is becoming more operational, more procedural, and more trust-conscious. The standout developments were not frontier-model breakthroughs. They were the expansion of business adoption support, a clinically meaningful health procurement step, more concrete public-service templates, and official cyber guidance for the frontier-AI era. (mbie.govt.nz)
That is a healthy sign of maturation, but it is only part of the story. New Zealand still appears to be moving faster on adoption support and guardrails than on research scale-up, sovereign compute execution, or breakout commercial capital formation. So the national picture in June 2026 is not stagnant, but neither is it explosive: momentum is real, broadening, and increasingly practical, while the biggest structural bets remain either unresolved or early. (mbie.govt.nz)