Voters want stronger action on AI. Can NZ afford to keep taking a soft approach?

As AI and its impacts become harder to avoid, other countries are setting rules for its use. New Zealand has some catching up to do.

Ads

Getty ImagesNext month, ChatGPT will mark its fourth birthday. In that short time, the popular chatbot and other generative artificial intelligence (GenAI) tools have transformed society as we know it.

Tech leaders have promised GenAI will unlock new levels of economic productivity and human prosperity. Critics warn it could instead deepen inequality. Whatever its long-term effects, the pressure to adopt it is growing.

New Zealand’s government expects agencies to use GenAI to create a leaner, more efficient public service. Companies are pushing “AI-first” initiatives and spending heavily on GenAI services to make sure they aren’t left behind.

Most New Zealanders have now had some interaction with this technology, with more than half of us using it at least weekly.

Some are trying to limit their use because of concerns about environmental costs or the impact on creative industries. But avoiding AI altogether is becoming increasingly difficult as it spreads through search engines, advertising and online content.

And while the current AI boom may eventually slow, GenAI is unlikely to disappear any time soon.

So how should New Zealand, a small nation far from the major economies hosting AI giants, learn to live with it? The next government has more choice than we might think.

Australia and other “middle powers” have already introduced controls to manage AI’s risks. With the election approaching, and parties pitching a range of policies, New Zealand is overdue a serious debate about what safeguards it needs.

NZ’s ‘light-touch’ approach

The government’s 2025 AI strategy deliberately takes a “light-touch approach”, encouraging AI adoption while relying largely on existing laws to manage its risks.

Rather than introducing specific AI regulations, it favours voluntary guidelines and frameworks that organisations are encouraged, but not required, to follow. This has raised concerns about whether legal protections can keep pace.

There is also public appetite for stronger action, with one survey finding eight in ten Kiwis believe AI regulation is needed. Another study found just 5% of respondents were strongly confident the government would use AI responsibly.

There are particular concerns about government agencies’ use of automated decision-making (ADM) systems, most of which are built on machine learning techniques that predate GenAI.

Done well, these systems can make life easier for people. ACC claims can be automatically approved before patients leave their GP’s office, while tax refunds can be paid without taxpayers having to take any action.

But done badly, it can make headlines. In 2018, Immigration New Zealand used an ADM model to identify overstayers for possible deportation, prompting accusations of racial profiling. In Australia, the Robodebt scheme issued unlawful welfare debts, forcing the government to repay A$721 million.

Read more: Election Policy Tracker: your guide to what the parties are promising during the 2026 election campaign

Yet New Zealand’s laws provide limited oversight of the technology. There are some protections under the Official Information and Privacy Acts, but they’re arguably unfit for purpose.

The latest GenAI systems use even more complex models, making it harder to identify problems such as bias and discrimination, which risks disproportionately affecting Māori and people with disabilities. Meanwhile, parliament recently passed a law allowing the Ministry of Social Development to automate welfare benefit decisions.

The Law Commission is presently developing a framework to govern ADM across government. Such efforts are overdue: Canada’s government began requiring its agencies to assess the impacts of such systems nearly seven years ago.

Who should benefit from AI?

Regulation is only part of the challenge. New Zealand also needs to consider who controls – and who benefits from – the AI systems it increasingly relies on.

Many of these systems, including those used for ADM, are developed and hosted overseas. This has implications for the protection of personal information, intellectual property and New Zealand’s ability to retain control over important public services.

Tech companies’ use of copyrighted material to train AI models has already prompted disputes over whether creators should be compensated.

There are other big questions about where the money goes. Much of New Zealand’s spending on AI flows to international technology companies, some of which declare relatively little income here. While many AI services are not yet profitable, their developers are betting on substantial future returns.

New Zealand needs to consider how those profits should be taxed before they materialise. The government might argue these giants are too difficult to tax. Critics would counter this simply reflects a lack of political leadership.

Arguments also extend to AI infrastructure. Data centres, such as one proposed for Southland, have received plenty of public attention. Their heavy energy demands and environmental impacts indeed deserve scrutiny, but they’re only one part of the debate about how New Zealand can benefit economically and socially from AI.

Other countries offer possible models for addressing these wider issues. Australia has established an AI Safety Institute, the European Union has introduced risk-based AI regulations and Canada requires government agencies to assess the impacts of automated decisions.

But New Zealand has its own obligations to consider and any approach must also reflect te Tiriti o Waitangi. That means considering Māori rights and interests and ensuring Māori involvement in designing and governing AI systems, particularly where Māori data is concerned.

Māori principles of data sovereignty emphasise community control over information and the benefits derived from it. As an example, the non-profit Te Hiku Media has already put these principles into practice, developing language technologies that keep control of te reo Māori data within the community.

These approaches show that governments and communities can have a greater say in how AI is developed and used. Whoever is elected to power next month, it’s clear New Zealand need not leave these decisions entirely to overseas technology companies.

From the cost of living to how best to tackle climate change, the coming election involves some big choices.

This article is part of The Conversation’s Election 2026 series: expert analysis of the major policy debates and the big challenges New Zealand’s next government will face.

Andrew Lensen receives research funding from the Ministry of Business, Innovation and Employment and the Royal Society of New Zealand. He is a co-director of LensenMcGavin AI, a consultancy specialising in the responsible uptake of AI in Aotearoa.

Tags: Academic Journalism, The Conversation, The Conversation Academic News