NZ risks losing control over how AI is used in critical sectors: expert
Monday, 16 March 2026
A Massey University researcher and PhD supervisor with a long background in computer science is sounding the alarm about New Zealand’s lack of standalone law, dedicated regulator or unified compliance regime for the use of Artificial Intelligence (AI).
Dr Athar Imtiaz, who is also AI and data lead at digital consultancy Nodero, said some of what he had seen in working with NGOs and others handling sensitive data shocked him, and the country must spend several hundred million its own framework for using the technology - or risk losing control over how it is used.
“If New Zealand does not define its own statutory and institutional expectations, the operating assumptions embedded in AI systems will increasingly be shaped elsewhere,” he said.
Imtiaz was sparked into looking closer at the country’s AI regulatory regime a few years ago when he was hired to write an AI policy for a local Primary Health Organisation (PHO), the name of which he cannot disclose.
Read more:
Iran war: Panic at the petrol station - Auckland Gull runs dry but not due to Middle East
High cost of power, consumer preference for cheap exports blamed for ‘devastating’ Wattie’s news
“They gave me their AI policy, and it was very funny in the sense it was just one A4 sheet of paper, and at the end of that, it said something to the effect that ‘people can use Copilot’ [Microsoft’s AI-powered conversational assistant integrated into the company’s apps, Windows systems, and browsers].
“It was pretty absurd given it was handling all sorts of PII [Personally Identifiable Information]. That was the extent of its policy.”
At that point, the AI expert started digging around to find what New Zealand did have for AI policy and guardrails, and found vanishingly little. The comparison with the European Union could not have been any greater. In mid-2024, the EU presented a world-first Artificial Intelligence Act (AI Act), legislating a detailed framework for regulating AI, banning things like linking people’s biometric data when writing policies for them, using facial recognition or scanning IT in the workplace (other than for health and safety reasons) and restricting things like using AI to evaluate credit ratings or the reliability of evidence.
In the time since, what Imtiaz called the EU’s “hefty document” has become heftier. The EU has now codified very strict rules around using AI across workplaces, and CV-scanning, education, critical infrastructure, and medical AI require strict risk management, high-quality data, and human oversight.
Just this month the EU rules went a step further, banning AI producing images of child sexual abuse in response to a stream of such imagery on social media platforms.
“I thought, ‘maybe New Zealand has an equivalent document or law’ - and I was double shocked to find we don’t have anything,” he told The Post. “That was very weird.” Weirder yet, he said, was that we didn’t follow an Australian set of rules - because they didn’t have much either.
While New Zealand does not have standalone AI law, it does have technology risks managed through the Algorithm Charter for Aotearoa - one of the world’s first all-of-government guides on AI use, the Privacy Act and general human rights protections.
While almost all government organisations and agencies are signed up to the Algorithm Charter, Imtiaz said it was not binding and merely set of recommendations.
“And they do have some tools which you can use and assess the state of AI governance and policy in your organisation. And they have a set of generic rules that, if followed, will lead to more ethical and proper use of AI, but again, it's not as detailed at all as compared to the European counterpart.”
Health systems
A very serious and urgent problem exists in the health system, where the use of AI tools and their capabilities have raced ahead of their governance.
One example Imtiaz offers is that of “Heidi”, an AI tool used by medical professionals pitched as something that “supports clinicians across the day from documentation to decisions and follow-up without adding burden”.
At the end of February Health Minister Simeon Brown said every emergency department in the country had access to the tool to help with things like generating draft clinical notes and writing referral letters.
And Health NZ is preparing to procure 1000 additional AI scribe licences for use by mental health teams.
Further advancements such as X-ray interpretation by AI and AI-based triaging is imminent, if not already being trialled, to help with a heavy workload and a stretched workforce. And that is as it should be, Imtiaz said.
“A lot of other countries are doing it and we need to too - but this is very sensitive data, and we need very strict rules around its handling.”
When AI assists with medical triage, welfare eligibility or justice processes, it becomes part of the decision-making infrastructure, and the technology generates likelihoods, rather than certainties.
“So we need clear standards for acceptable error and bias testing, and validation against datasets that properly reflect New Zealand’s communities, including Māori, rather than relying solely on international training data.”
Investment
Building sovereign capability would require investment in high-performance computing, secure data environments and specialist expertise - a system that could require several hundred million dollars over the next five years.
Other countries are doing it. Australia’s committed more than A$100 million in federal funding toward AI capability and regulatory reform; Singapore has invested billions across successive national AI strategies and the UK funds AI research and governance capacity through its central technology portfolio.
Imtiaz said New Zealand had attracted investment from global cloud providers expanding local data centre capacity. But that did not equate to sovereign control.
“We have the opportunity to design a framework that reflects New Zealand’s legal system, social expectations and Treaty commitments.
“In the upcoming election, every party should have thought about AI and have smart policies to present to voters about how the intend enshrine sovereignty over the use of this technology.”