Atamai hangahanga me ngā mauhanga tūmatanui
Artificial intelligence and public and local authority records
Learn how your public sector organisation can continue creating and maintaining full, accurate and accessible records when using artificial intelligence.
The document identifier for this guidance is 23/G25 and it was approved in December 2023.
Please note that there is no PDF version of this page because the content is the same and our website has a feature to print a webpage. You can also see a preview of this by using the print command (CTRL + P). If you still require a PDF version, please contact us at rkadvice@dia.govt.nz
Who this guidance is for
We’ve designed this guide to help public offices and local authorities (public sector organisations) continue to create and maintain full, accurate and accessible records when adopting and using artificial intelligence (AI).
You can use this guidance to identify risks and considerations for information and records created or affected by using AI. It also:
supports meeting the requirements of the Public Records Act 2005 (the Act) and Information and records management standard (the Standard)
aligns with the public sector system leaders’ guidance for use of generative AI (Gen-AI).
What artificial intelligence is
AI refers to computer systems that do tasks which seem like intelligent behaviour — such as finding patterns or categorising. This includes generative AI tools and other systems that interpret data or automate tasks. Some specific examples of AI are outlined further down this page.
How artificial intelligence relates to public and local authority records
Outputs created using AI are public and local authority records
Information recording the adoption and use of AI related to public and local authority records is subject to the Act and the Standard. The outputs created using AI are public and local authority records.
AI technologies can be used to inform decisions and carry out key business activities. The way AI is used — and the outputs — are integral components of full, accurate and accessible records of the affairs of government.
AI’s impact on information and records should be documented
To be transparent and accountable, it’s important to accurately record how and why public sector organisations’ actions and decisions are made.
You should document the use of AI and its effect on information and records. This is because the processes and technologies used impact how information and records are:
formed
captured, and
understood in context.
This ensures their integrity as evidence and their accessibility now and in future.
How you can support trusted use of artificial intelligence
You can support trusted use of AI by:
using the system leaders’ guidance for adopting trusted use of AI for the public service, including their interim generative AI guidance for the public service and 10 do’s for trustworthy use of GenAI for the public service
applying the Privacy Commissioners recommendations for responsible use of AI
using the Algorithm charter for Aotearoa New Zealand and applying the guidance
using the Information and records AI checklist.
Information and records AI checklist
How we developed the checklist
This Information and records AI checklist (Word, 112KB) is based on a process developed by Te Whatu Ora Health New Zealand in 2023 that reflects good practice. You can find out more about Te Whatu Ora and their work on implementing an AI governance framework in the article linked below.
Read ‘An example of governance for AI in health services from Aotearoa New Zealand’
How the checklist can help you
The Information and records AI checklist:
aligns with the system leaders’ guidance and
aligns with our recommendations for compliance with the Act and the Standard, and
supports good decision making in public sector organisations about AI adoption and use.
The checklist is based on lifecycle stages across adoption decisions and operational use. It will help you manage the development and adoption of new AI for your organisation. You can adapt and edit the checklist to meet your specific business needs.
Although all aspects of the Act and the Standard may apply to public and local authority records affected by using AI, some areas are highlighted in the checklist. This is to help you address key questions specific to the Act and the Standard, alongside other good practice considerations.
The checklist helps ensure transparency, accuracy and accountability by recording:
what the AI application or system is
what the AI application or system will be used for
how it will be used.
Examples of good practice and AI use
Public Record Office of Victoria in Australia guidance
International examples of good practice feature in guidance developed by the Public Record Office of Victoria in Australia. This includes:
why it’s important to capture records of AI technologies
capturing records of AI use
documenting AI
international sources and published research on trusted use of AI.
View the Public Record Office of Victoria guidance on their website
The State of AI Governance in Australia report
The State of AI Governance in Australia report — published in 2023 by the University of Technology Sydney’s Human Technology Institute — includes examples of AI and how it is used.
Machine learning systems
Machine learning systems are a broad set of models that have been trained on preexisting data to produce useful outputs on new data.
Expert systems
Expert systems are systems that use a knowledge base, inference engine and logic to mimic how humans make decisions.
Natural language systems
Natural language systems are models that can understand and use natural language and speech for tasks such as summarisation, translation, or content moderation.
Facial recognition technologies
systems that verify a person, identify someone, or analyse personal characteristics using face data drawn from photos or video.
Recommender systems
Recommender systems suggest products, services or information to a user based on user preferences, characteristics, or behaviour.
Automated decision-making systems
Automated decision-making systems use data to classify, analyse and make decisions that affect people with little or no human intervention.
Robotic process automation
Robotic process automation are systems that imitate human actions to automate routine tasks through existing digital interfaces.
Virtual agents and chatbots
Virtual agents and chatbots are digital systems that engage with customers or employees via text or speech.
Generative AI
Generative AI are systems that produce code, text, music, or images based on text or other inputs.
AI-powered robotics
AI-powered robotics are physical systems that use computer vision and machine learning models to move and execute tasks in dynamic environments.