What exactly will Labour’s Office of AI actually be allowed to do?

Labour launched its ‘AI in New Zealand’s Interests Action Plan’ on 9 September 2026, and for businesses weighing up AI investment the problem is not what it promises. It is what it leaves out. As Newsroom put it in its analysis of the launch, the policy is silent on how any of these rules will be enforced, what powers the new officer will have, or how fines will be doled out. Labour’s answer is that those details get sorted after the election. That is not a framework a compliance team can plan around.

What Labour is actually promising

The suite is substantial on paper. Labour would set up an Office of AI within the Department of Prime Minister and Cabinet, using existing resources, to coordinate policy across government and develop standards, with a senior minister carrying AI responsibility. Labour leader Chris Hipkins said he wanted it running certainly within the first few months of taking office.

Alongside it: a broadened online safety regulator inside Internal Affairs, a copyright framework so creators are paid when their work trains AI models, and new rules requiring data centres to secure their own renewable energy supply and pay for their own grid connections. Add curbs on automated decision-making in the public service and a ban on non-consensual deepfake pornography, and Labour has clearly decided AI is a live election issue.

Hipkins framed it as a sovereignty argument. Used well, he said, AI can cut hospital paperwork and make small businesses more productive, but “we cannot leave it to big international tech companies to decide how it shapes our future”. On automated decisions he was sharper still, saying New Zealanders shouldn’t have decisions made about their lives by an algorithm “that no one can explain and no one is accountable for”.

The questions a compliance team would ask first

Here is where the policy thins out. The document does not say what statutory powers the Office of AI would hold, how fines would be calculated or collected, what mechanism would underpin the copyright rules, or how the data centre energy conditions would be monitored and penalised. Science spokesperson Reuben Davidson said Labour would build the copyright rules “with creators, publishers and the tech companies” and “make sure they can actually be enforced”. Fine, but enforceability is precisely the thing not yet designed.

For a business owner, regulatory uncertainty is itself a cost. You cannot budget for a compliance regime whose shape, powers and penalties are all ‘to be confirmed’. An Office of AI stood up inside DPMC ‘using existing resources’ with no fresh statutory teeth is a coordination body, not a regulator. Calling it one before the powers exist is where the policy overreaches.

The backdrop that should worry both sides

The current government took the opposite approach in its July 2025 AI Strategy: light-touch, principles-based, leaning on existing privacy and consumer law rather than standalone AI legislation. That has done little for smaller firms. As of a 2024 NZIER and Spark survey cited in the strategy, 68% of NZ SMEs had no plans to evaluate or invest in AI, against just 38% of Australian SMEs. Larger firms are moving, with a 2024 Datacom survey showing 67% now using some form of AI, up from 48% in 2023.

The international scoreboard is unflattering. MBIE’s August 2025 companion document put New Zealand 40th of 188 countries on the Oxford Insights AI readiness index and 35th of 36 on the Stanford AI vibrancy indicator. Meanwhile the EU has its AI Act, the UK its Digital Markets framework, and the US FTC its rule-making, as the Commerce Commission catalogued in July 2025. New Zealand has none.

The prize that makes getting this right urgent

The stakes justify the debate. Accenture and Microsoft estimates cited in the same 2025 MBIE document suggest generative AI could add $75 billion in value to the economy by 2038, lift GDP 15% above baseline and free up 275 hours per worker annually. Get regulation too heavy and you chill that investment. Too vague, and you create the compliance limbo Labour is now offering.

What neither party has is a plan to bring smaller firms along. Commentator Peter Dunne noted in August 2026 that Taiwan committed NZ$5.33 billion specifically to accelerate AI in its SME sector, concluding that “no party has committed to upskilling small businesses to work effectively in the AI environment, the way Taiwan is.” Labour’s plan aims at the tech giants and leaves the two-thirds of Kiwi SMEs sitting on the sidelines exactly where they are. The regulator can wait for its powers. The adoption gap cannot.

Sources

• Labour rides public wave to dump on AI (2026-09-09)
• Labour promises to set up AI regulator, copyright rules if elected (2026-09-09)
• Labour pledges AI office, data centre rules, automated decision curbs (2026-09-09)
• Labour promises new rules for data centre electricity use, coordinated AI policy across government (2026-09-09)
• Labour plans ‘Office of AI’ and wider role for government in its adoption (2026-09-08)
• New Zealand’s AI Strategy: Investing with Confidence (2025-07)
• Navigating the Rise of AI (2025-07-07)
• AI? No can do. NZ politics have become myopic (2026-08-20)