New Zealand's approach to AI governance in 2023 can be summarised as: "we're watching, we're learning, and we're not ready to regulate yet." Given the pace of AI development, that's both pragmatic and slightly concerning.
What's working:
NZ is following the OECD AI Principles, a sensible baseline that covers fairness, transparency, accountability, and human oversight. Rather than rushing new legislation, we're leaning on existing frameworks: the Privacy Act, the Public Records Act, and sector-specific regulation. This avoids the risk of premature laws that don't fit the technology.
The Digital Government team has begun developing guidance for public sector AI use. It's early-stage, but the direction is right: practical frameworks rather than theoretical policy.
What's missing:
Clear guidance for enterprise. The government has signalled support for AI adoption but hasn't provided practical guidance for private sector governance. Enterprises are left to self-regulate, which produces inconsistent approaches and governance gaps. A national AI governance framework, even a voluntary one, would give enterprises a credible baseline to build on.
Workforce transition planning. AI will reshape work across industries, and NZ's small, concentrated economy means those shifts will be felt acutely. There's minimal public discussion about how to manage the transition: reskilling, role redesign, new job creation. This gap will become urgent in 2024-2025.
Data sovereignty clarity. Where can NZ enterprise data be processed by AI? What are the boundaries? For government and regulated industries, this matters enormously. The current guidance is vague enough to create uncertainty without creating safety.
Māori data sovereignty integration. Te Mana Raraunga (the Māori Data Sovereignty Network) has articulated clear principles about Māori data governance. These principles should be explicitly integrated into NZ's AI governance framework, not treated as a separate consideration. AI systems processing data about Māori communities need governance that respects indigenous data sovereignty. This isn't optional, and it's not yet embedded in the national approach.
My assessment: NZ gets a B- for AI governance in 2023. The foundations are sensible. The pace is too slow. The private sector is moving faster than the public sector guidance can keep up with. Enterprise leaders can't wait for government frameworks. They need to build their own governance approaches now and adapt when national guidance arrives.
The good news: NZ is small enough that a determined effort in 2024 could close these gaps quickly. The question is whether there's political will to prioritise it.