Overview of the Public Service AI Framework

The Public Service AI Framework:

• sets a vision for AI in New Zealand’s Public Service
• defines 5 key principles to guide Public Service AI use
• outlines the policy context for Public Service AI use
• outlines 6 key pillars of the Government Chief Digital Officer’s (GCDO) AI work programme.

The Public Service AI Framework provides support for agencies in their use of AI, and how to implement these technologies safely. Agencies are encouraged to align with the direction set by this Framework, however it’s not binding.

For practical examples of how the Framework might guide the deployment of AI within your organisation, see Applying the Framework.

Intended outcome of the Framework

The AI Framework supports the overall outcome that:

The Public Service models best practice in AI use, enabling and contributing to the wider community and economy in alignment with New Zealand’s position as a trusted global partner.

Why do we need a Framework?

Harness the potential of AI for the public service

In June 2024, Cabinet set a strategic direction for New Zealand’s use of AI. The potential for harnessing AI to enhance customer experience and boost efficiency is significant. The Government is seeking ways to enable safe AI innovation in public services that support better outcomes for all New Zealanders.

Cabinet Paper — Approach to work on AI — MBIE (PDF 155KB)

Manage the risks of AI for the public service

AI is evolving rapidly and if inaccuracy or misuse were to cause failure in public service delivery, the consequences could be serious. Public services are rightly held to a high standard, and privacy, security and ethical risks need to be carefully considered and well managed.

It’s critical that agencies understand and make informed decisions about these risks to maintain public trust and protect the integrity of their systems and the Public Service as a whole.

Scope and audience of the Framework

The Public Service AI Framework has been developed to:

• support a structured approach to the development, deployment and use of AI across the New Zealand Public Service
• support leaders, decision-makers, practitioners and influencers of AI within Public Service agencies to use AI lawfully and in line with Public Service values.

Values — Public Service Commission

What kinds of AI does the Framework apply to?

Defining AI

The OECD defines an AI system as:

a machine-based system that, for explicit or implicit objectives, infers, from the input it receives, how to generate outputs such as predictions, content, recommendations, or decisions that can influence physical or virtual environments. Different AI systems vary in their levels of autonomy and adaptiveness after deployment.

AI is technology that enables computers and machines to simulate human learning, comprehension, problem solving, decision making, creativity and autonomy.

The Public Service AI Framework applies to all forms of AI used in New Zealand public services.

Vision of the Framework

Adopt AI responsibly to modernise public services and deliver better outcomes for all New Zealanders.

The Framework’s vision sets the direction for the safe and responsible use of AI to enable more efficient, responsive and high-performing Public Services.

It encourages agencies to pursue using AI to create value for New Zealanders, while managing risks and maintaining public trust and confidence in the Public Service overall. It also encourages a human-centric approach of AI and other emerging technologies for public services. This will ensure citizens, customers, taxpayers and public service workers are at the forefront of the design and implementation of Artificial Intelligence.

The 5 principles of the Framework

We use the OECD’s values-based AI principles to inform the principles of the Public Service AI Framework.

AI principles — OECD

Inclusive, sustainable development

Public Service AI systems should contribute to inclusive growth and sustainable development through a focus on innovation, efficiency and resilience, and on reducing economic, social, gender and other inequalities and protecting natural environments. AI use should consider and address concerns about unequal access to technology.

Human-centred values

Public Service AI use should respect the rule of law, democratic values and human rights and labour rights through the lifecycle of each AI system or product.

These rights and laws include personal data protection and privacy, dignity, non discrimination and equality, self determination and autonomy. Public service workers have the right to be consulted on changes made to their work and working arrangements. Agencies need to provide human oversight throughout the AI lifecycle to ensure ethical and appropriate use.

Transparency and explainability

The Public Service needs to commit to transparency in its use of AI. People interacting with government AI systems or receiving AI-assisted services should be aware of and understand how AI is being used.

To support this, agencies should publicly disclose:

• when AI systems are used
• how they were developed
• how they affect outcomes — as relevant and appropriate according to the given use case.

Agencies should also enable people affected by the outcome of an AI system to understand how the outcome was determined.

Safety and security

Public Service AI systems should treat the security of customers and staff as a core business requirement, not just a technical feature (security-by-design). They should minimise risk to individual or national safety and security under normal use, misuse or adverse conditions.

Security-by-design — National Cyber Security Centre

The Public Service should ensure traceability of data, apply a robust risk management approach and work collaboratively with commercial and security colleagues in the procurement and assurance of AI tools.

Accountability

AI use within the Public Service should be subject to oversight by accountable humans with appropriate authority and capability at every stage.

This should include the application of relevant regulatory and governance frameworks, reporting, auditing and/or independent reviews.

Agency AI capabilities need to keep pace with technological changes, to maintain a strong understanding of AI systems and their limitations.

Background to the Framework principles

The OECD AI principles

Cabinet agreed in June 2024 to promote the OECD principles as a key direction for our approach to responsible AI in New Zealand. The OECD principles promote the use of AI that’s innovative and trustworthy and that respects human rights and democratic values.

Cabinet Paper — Approach to work on AI — MBIE (PDF 155KB)

OECD AI Principles overview — OECD.AI Policy Observatory

Other guiding frameworks and principles

The Framework’s principles are also informed by:

• UK’s Generative AI Framework — UK Government
• Algorithm charter for Aotearoa NZ — Data.govt.nz
• The AI Principles — AI Forum NZ.

New Zealand policy context

AI laws, regulations and conventions

Government agencies need to use AI in ways that are in line with the existing laws, regulations, conventions, policies and guidance that govern or have application for public service AI use. The AI Framework recognises some key areas within this wider policy context, including the:

• Treaty of Waitangi
• Official Information Act 1982
• Bill of Rights Act 1990
• Human Rights Act 1993
• Copyright Act 1994
• Public Records Act 2005
• Public Service Act 2020
• Privacy Act 2020.

New Zealand’s approach to AI regulation

Cabinet has agreed to a light-touch, proportionate and risk-based approach to AI regulation. They’ve recognised the NZ policy context already provides some guardrails and have agreed that further regulatory intervention should:

• only be considered to unlock innovation or address acute risks
• use existing mechanisms in preference to developing a standalone AI Act.

Cabinet Paper — Approach to work on AI — MBIE (PDF 155KB)

GCDO Public Service AI Work Programme

The GCDO is leading a Public Service AI work programme to support the implementation of the Framework’s vision. The work programme will iterate and adapt to changes in context and technology, and support the Public Service to use AI safely and responsibly.

The work programme’s 6 pillars

The following 6 pillars guide the focus of the GCDO-led Public Service AI work programme:

• Governance — to support transparency and human accountability in Public Service AI use.
• Guardrails — to enable safe and responsible Public Service AI use.
• Capability — to build internal and external AI knowledge and skills.
• Innovation — to provide pathways to enable safe AI testing and innovation in the Public Service.
• Social licence — to ensure New Zealanders and public service workers have trust and confidence in Public Service AI use.
• Global voice — to ensure international counterparts see New Zealand as a trusted AI partner.

The GCDO is working closely with MBIE which is compiling a cross-portfolio policy work programme and National AI Strategy.

Applying the Framework

The Public Service AI Framework supports government agencies to ensure that they’re using AI lawfully, safely and responsibly for its benefits, while maintaining public trust and aligning with overall Public Service values. The Framework also supports a human-centric approach of AI and other emerging technologies. This will ensure citizens, customers, taxpayers and public service workers at are at the forefront of the design and implementation of Artificial Intelligence and outcomes of this change for both the public and public service workers are positive and equitable.

We recommend that agencies refer to and use the Public Service AI Framework to guide the responsible deployment of AI within their operations.

This could include:

• Guiding AI development — referring to the Framework and ensuring alignment with its principles during design and implementation of AI initiatives
• Policy formulation — using the Framework to inform the development of organisational AI governance policies and guardrails for the use of AI
• Skills building — using the Framework to inform the development of training programmes for users, to ensure they understand how to use AI in line with organisational and public service values and rules
• AI evaluation — using the Framework to assess AI initiatives against its key values, principles and rules.

Contact information

For further information please email the GCDO team: gcdo@dia.govt.nz