AI in New Zealand: the 2026 landscape.

The Government’s position: adopt, don’t legislate

New Zealand published its first national AI strategy in July 2025. As MBIE put it: “Until now, New Zealand was the only OECD country that had not published an AI strategy”. The strategy is about using AI, not building it. It “deliberately emphasises AI adoption and application rather than foundational AI development”, and states: “New Zealand is taking a light-touch and principles-based approach to AI policy.” Risks are left to existing law on privacy, consumer protection and human rights.

Quote the exact document you rely on. The 2024 Cabinet paper said “light-touch, proportionate and risk-based”; the formal Cabinet minute records only a “proportionate and risk-based approach to AI regulation where needed”. In June 2025 Cabinet added a commitment to keep New Zealand’s rules “no more restrictive than comparator countries in the OECD”.

This is the current Government’s stance, dated 19 October 2026. RNZ reported that National was sticking with the strategy, while Labour’s plan “promises to establish an Office of AI within the Department of the Prime Minister and Cabinet”. Recheck after the election.

In government, the 2026 cross-agency survey found “59 organisations reported 545 AI use cases”, double 2025’s 272, with 31% operational.

Privacy law: the rules that already bite

The Privacy Commissioner’s position is the starting point for any AI project: “Our starting point is that the Privacy Act applies to everyone using AI tools in New Zealand.” Four developments matter in 2026.

• IPP 3A is in force. Since 1 May 2026, an agency collecting personal information “from a source other than the individual themselves must take reasonable steps to ensure that the individual is aware of specified matters.” There are now 14 information privacy principles, and IPP 3A does not apply to information collected before that date. AI tools that enrich records from the web are a common trigger.
• OPC expects a privacy impact assessment before AI tools are used, along with human review before acting on outputs.
• Biometrics have their own code. The Biometric Processing Privacy Code 2025 applied to new processing from 3 November 2025, and the transition period for existing users “ended on 3 August 2026”.
• Penalties remain light. The Commissioner says “in New Zealand there’s no civil penalty regime”, and has called for stronger protections around automated decision-making. Failing to notify the Commissioner of a notifiable privacy breach carries a maximum fine of $10,000.

Offshore AI vendors are covered too: the Act applies to overseas agencies carrying on business here. And the public is wary. In OPC’s 2026 survey of 1,245 people, 67% were concerned about agencies and businesses using AI to make decisions about people from personal data. Our lesson on protecting customer information covers the day-to-day practice.

Sector regulators are setting the practical rules

With no AI Act, sector bodies are writing the operative expectations.

Two policy tracks are open. Cabinet wants a report back by 31 March 2027 “on a possible copyright framework for GenAI in New Zealand”, and a member’s bill extending intimate-image offences to deepfakes awaits a select committee report due 20 November 2026.

Infrastructure: where AI models actually run

A cloud region in New Zealand means data can be stored here; it does not mean an AI model processes it here. As at 19 October 2026, no major AI vendor checked offers guaranteed model processing inside New Zealand.

Physical capacity is coming. Datagrid has begun site works on a $3.5 billion AI data centre campus at Makarewa, north of Invercargill, designed for “high-density GPU clusters serving global AI and cloud providers”. For architecture choices, see our guide to data residency for AI systems.

The ecosystem: funding, institutions and a cautionary tale

• AI Forum New Zealand, founded in 2017, reports more than 280 members and runs the free AIGovernance.nz toolkit.
• NZ Institute for Advanced Technology (NZIAT): five AI research platform concepts were shortlisted in December 2025, each with $250,000 to develop a proposal. A final platform was expected in the first half of 2026; we found no announcement of a winner by 19 October 2026.
• Callaghan Innovation is “being disestablished”, not closed: it is still operating in 2026/27, while R&D grants and Research and Development Tax Incentive support have moved to MBIE.
• Te Hiku Media leads Papa Reo, a te reo Māori language-technology programme it says was awarded $13 million, and publishes its own Kaitiakitanga License, which bars commercial use of its code without explicit permission.
• Soul Machines, the Auckland digital-human company, went into receivership on 5 February 2026; NBR reported it owed creditors more than $19.5 million.

How many organisations use AI: only vendor surveys can say

There is no official figure. Stats NZ’s Business Operations Survey “was last carried out in 2023, and was paused for 2024 and 2025”. Its successor, the 2026 Survey of Business Operations, asks about AI, but no results had been published by 19 October 2026. Every adoption percentage in circulation comes from a commercial survey, and they ask different people.

Depth is thinner than reach. Datacom found only 15% of larger organisations scaling AI organisation-wide, and 64% of leaders wanting specific AI legislation, a contrast with the Government’s stance. Public trust is low: in the KPMG and University of Melbourne global study, “Only 44% of New Zealanders believe the benefits of AI outweigh the risks, the lowest ranking of any country.”

What this means for a New Zealand organisation

The absence of an AI Act does not mean an absence of obligations. Take legal advice for your situation.

• Run a privacy impact assessment before AI touches personal information
• Check whether each vendor uses your data for its own purposes, such as training
• Send IPP 3A notices when AI enriches records from other sources
• Record where models process data, not only where it is stored
• Substantiate every claim that your product uses AI
• Read your sector regulator’s AI guidance and date your review
• Consult staff before AI changes that affect their roles
• Apply the Te Mana Raraunga principles wherever Māori data is involved

Our AI governance advisory turns this into policy and controls, and our AI use policy template is a free place to start. For regulated work, see AI for financial services or talk to us about AI consulting in New Zealand.

What to watch next

• First official adoption figures from the 2026 Survey of Business Operations.
• The 2026 general election and whether the strategy survives.
• The deepfake bill, with a select committee report due 20 November 2026.
• The NZIAT AI research platform decision.
• The generative AI copyright report-back, due 31 March 2027.

Sources were read on 19 October 2026. Policy is moving quickly, so check dates before relying on any line.

Written by the Levy Street team in Wellington. Last updated 19 October 2026.