AI Regulation Tracker / Frameworks and guidance

New Zealand's Public Service AI Framework: Non-Binding Guidance

In July 2025, alongside its National AI Strategy, New Zealand published a Public Service AI Framework through the Government Chief Digital Officer. It sets six pillars for how public agencies should use AI, including transparency about where AI is in the mix. The Framework is a signal of direction, not a binding rule. In its own words, agencies "are encouraged to align with the direction set by this Framework, however it's not binding."

By Anthony Guerriero · The Leveraged Years Regulatory Research · Published July 2025 · Last updated July 2025 · Framework analysis

The short version

New Zealand's Government Chief Digital Officer published a Public Service AI Framework in July 2025, sitting inside the wider National AI Strategy. It groups the government's AI work into six pillars: Governance, Guardrails, Capability, Innovation, Social licence, and transparency threaded through the Governance pillar. It encourages agencies to be transparent and explain how they use AI in their work. The important point for anyone reading this as a compliance item is that the Framework is voluntary. The page states plainly that agencies "are encouraged to align" but that it "is not binding." Cabinet has picked a light touch, risk based approach to AI, and New Zealand has no dedicated AI statute. My read: treat this as a preview of where a future rulebook may point, and as a reasonable governance checklist to borrow, rather than a legal obligation you can breach. Primary source: digital.govt.nz Public Service AI Framework, plus the MBIE strategy paper Investing with confidence.

Regulatory Briefing

Instrument: Public Service Artificial Intelligence Framework, within the National AI Strategy ("Investing with confidence")
Authority: Government Chief Digital Officer (GCDO), Department of Internal Affairs, New Zealand; strategy work led with the Ministry of Business, Innovation and Employment (MBIE)
Jurisdiction: New Zealand (central government / public service)
Status: Published July 2025, active guidance
Bindingness: Non-Binding Framework (SIGNAL). The page states agencies are encouraged to align but that the Framework is not binding. It does not carry penalties and is not a statute.
Relevant date: Launched July 2025 with the National AI Strategy; no compliance deadline attaches, because alignment is voluntary
Primary Source: digital.govt.nz, Public Service AI Framework, digital.govt.nz
Workflow Impact: Sets an expectation that public agencies document and disclose AI use and keep human accountability, which shapes how vendors and contractors present AI tools to NZ government buyers even though the Framework binds no one.
Operational Next Step: If you sell to or work inside NZ public bodies, map your AI use against the six pillars now, so you are ready if any of this direction later moves into rules.
Last Verified: July 23, 2026

Here is the setup. In July 2025 New Zealand launched a National AI Strategy branded "Investing with confidence." Inside it, the Government Chief Digital Officer published a Public Service AI Framework aimed at the central government's own use of AI. The Framework groups the work into six pillars and points agencies toward safe and accountable adoption. It is guidance, not law, and the page says so directly.

New Zealand has taken a lighter path than the European Union. Cabinet agreed to what it calls a light touch, proportionate and risk based approach, and it has said further regulatory intervention should only be considered to open room for innovation or to address acute risks. There is no dedicated AI Act in force. The Framework carries the government's direction of travel, and it is built on the OECD AI principles that Cabinet agreed to promote in June 2024.

What are the six pillars?

The Framework organizes the public service AI work programme under six headings. Governance is there to support transparency and human accountability. Guardrails are meant to keep use safe and responsible. Capability covers building AI knowledge and skills inside and outside government. Innovation is about giving agencies safe ways to test AI. Social licence is about keeping the trust of New Zealanders and public service workers. Transparency and explainability run through the whole set as a principle, which is where the disclosure expectation lives.

Read together, these are the same categories a careful private organization would use for its own AI governance. That is part of why the document is useful beyond Wellington. It is a compact, government backed checklist that happens to be free to borrow.

Does the Framework require anyone to disclose AI use?

No. This is the line I want to be precise about, because it is easy to overstate. The Framework encourages agencies to be transparent about their AI use and to keep human accountability, and transparency is one of its stated principles. What it does not do is compel anything. The page is explicit that agencies "are encouraged to align with the direction set by this Framework, however it's not binding." There is no deadline, no penalty, and no regulator standing behind it with an enforcement power. Calling this a disclosure rule would be wrong. It is a disclosure expectation set by guidance.

That distinction matters in practice. A voluntary framework shapes behavior through procurement, reputation, and the risk that today's guidance becomes tomorrow's requirement. It does not give anyone a cause of action, and an agency that falls short of it is not breaking a law.

Why does a voluntary New Zealand framework reach US professionals?

Two reasons, and neither depends on the Framework binding anyone in the United States, because it does not.

First, if your firm sells AI tools or AI enabled services to New Zealand government buyers, this document tells you how those buyers are being told to think. A procurement team that has internalized six pillars and a transparency principle will ask you where the AI sits in your product, how it was built, and how a human stays accountable for the output. You will answer those questions better if you have read the Framework than if you meet it cold in a tender.

Second, this is a clean comparative reference for anyone building an internal AI governance policy anywhere. It shows a national government settling on a small, legible set of principles rather than a long rulebook, and choosing to lead with encouragement while it watches how the technology develops. For a US executive weighing how heavy to make an internal AI policy, New Zealand's choice is a data point worth having.

What should you actually do with it?

If you touch the New Zealand public sector, line your own AI use up against the six pillars and be ready to speak to each one, especially governance and transparency, since those are where a buyer's questions will start. If you are drafting internal AI governance for a private organization, treat the Framework as a template you can adapt, not a standard you have to meet. And keep an eye on the wider strategy work MBIE is running, because a voluntary framework is often the first draft of something that later carries more weight.

Questions professionals are asking

Is New Zealand's Public Service AI Framework a law?

No. It is voluntary guidance published by the Government Chief Digital Officer within the July 2025 National AI Strategy. The Framework page states that agencies are encouraged to align but that it is not binding. New Zealand has no dedicated AI statute, and Cabinet has chosen a light touch, risk based approach.

Does the Framework require agencies to disclose their AI use?

Not in a legal sense. Transparency is one of the Framework's principles, and it encourages agencies to be transparent about how they use AI while keeping human accountability. Because the Framework is not binding, that transparency is an expectation set by guidance, with no deadline or penalty behind it.

What are the six pillars?

Governance, Guardrails, Capability, Innovation, and Social licence, with transparency and explainability running through the set as a principle. The pillars organize the Government Chief Digital Officer's public service AI work programme.

Does this affect anyone outside New Zealand?

Not as law. It binds only New Zealand's public service, and only as encouragement. It is useful to US readers as a guide to how NZ government buyers are told to think about AI, and as a compact governance template that any organization can adapt.

RELATED BRIEFINGS

Browse the full AI Regulation News tracker

Informational analysis for working professionals, not legal advice. Confirm how any framework, statute, or requirement applies to your situation with qualified professionals in the relevant jurisdiction.