Responsible AI in Action
This guidance from the New Zealand Ministry for Regulation provides practical steps for regulatory leaders and practitioners on adopting and using artificial intelligence responsibly within regulatory systems. It covers foundational requirements such as strategy, governance, procurement, and workforce readiness, as well as operational considerations like human-in-the-loop oversight and ethical boundaries. Real-world case examples from the Department of Conservation and Hutt City Council illustrate how AI can support regulatory tasks while maintaining human accountability.
Summary and content fields are drawn from the document text and are pending editorial review.
Who issued it
Who it governs
Senior leaders and management teams in regulatory organisations who are responsible for regulatory systems and statutory decision-making.
What the document contains
This document addresses how regulatory agencies can introduce artificial intelligence into their operations safely and ethically. It covers foundational strategy, vendor engagement, workforce readiness, practical use cases, human-in-the-loop safeguards, iterative piloting, and adherence to public sector ethics and Te Tiriti o Waitangi obligations.
The document does not provide binding legal mandates or statutory penalties for non-compliance, nor does it establish technical certification standards for AI algorithms.
- AI adoption in regulatory agencies should be treated as a strategic regulatory capability rather than a standalone IT project.
- Accountability for regulatory decisions always remains with human decision-makers, supported by effective human-in-the-loop safeguards.
- Regulatory leaders should start small with low-risk, reversible pilots to build staff confidence and organisational capability.
- Robust procurement practices are essential to manage vendor risk, protect data privacy, and prevent vendor lock-in.
- Te Tiriti o Waitangi obligations and Māori data sovereignty principles must be integrated into AI-enabled regulatory practices.
· Why this? Why now? · Section one: set the foundations · Section two: use AI well (and keep humans in control) · Section three: navigating AI ethically · From guidance to practice: AI checklist
How it scores
Each score is backed by a quote verified against the document text. Editorial review has not yet confirmed them.
Accountability
4/5Clear responsibility for AI outcomes
None of this transfers responsibility away from your organisation. Accountability remains with the legally authorised decision-makers, not with the tool and not with the vendor.
Section one: set the foundations
screening score · 2026-10-23
Fairness
4/5Non-discrimination and equitable treatment
Regulators need to be alert to these risks and treat fairness as an ongoing responsibility. This includes: • understanding, at a high level, how AI systems work and where their limitations lie
3.1 Setting ethical boundaries
screening score · 2026-10-23
Human Oversight
4/5Humans remain in control of consequential decisions
In regulatory settings it is also important to design for meaningful human oversight in decision-making, and to ensure people can request review, including an appeal to a human where AI has influenced an outcome.
1.1 Start strong: strategy, governance, and guardrails
screening score · 2026-10-23
Privacy
3/5Protection of personal data and private life
Privacy risks also increase when AI draws on large or sensitive datasets. Regulators must apply strong data governance and ensure personal information is handled lawfully, securely, and proportionately.
3.1 Setting ethical boundaries
screening score · 2026-10-23
Transparency
4/5Openness about how AI systems work and are used
Regulators already have a duty to explain how and why decisions are made. That duty does not change when AI supports or informs those decisions.
3.1 Setting ethical boundaries
screening score · 2026-10-23
Enforcement
4/5Consequences and compliance mechanisms with real force
Include clear clauses on: • performance measures and audit rights • security standards • ongoing monitoring and reporting • lifecycle management, including updates and decommissioning
1.2 Procure with purpose: AI vendor engagement
screening score · 2026-10-23