Manage risk and maximise opportunity with first of its kind AI system management standard ISO/IEC 42001
With the rapid rise in Artificial Intelligence weaving its way into many facets of digital life, it’s timely that a new ISO/IEC standard developed by international committees will help people navigate safe and responsible use.
With AI rapidly integrating into daily life, a new international standard—ISO/IEC 42001:2023—has been developed to guide its safe, ethical, and responsible use. This milestone comes as artificial intelligence shapes various digital interactions, from e-commerce chatbots and smart home devices to speech recognition, navigation, healthcare, and streaming services. AI is no longer a future concept; it is already embedded in the fabric of everyday life.
As AI applications collect and use data, they must comply with global regulations. The ISO/IEC 42001:2023 standard provides a framework for businesses to manage AI safely across multiple use cases. Craig Pattison, Chief Operating Officer of the Capability Collective and agricultural stream lead for the AI Blueprint for Aotearoa 2024, highlights its importance:
‘It is the first of its kind and the international standard for establishing, implementing, maintaining, and continually improving an Artificial Intelligence Management System (AIMS). This standard is designed to ensure that AI systems are developed and deployed responsibly, ethically, and securely.’
New Zealand’s contribution to AI Standards
New Zealand is one of over 60 countries contributing subject matter expertise to ISO/IEC JTC 1/SC 42, the international AI committee. The New Zealand mirror-committee, chaired by Craig Pattison, represents the country’s cultural and business needs in this "one country, one vote" process.
Recognising that businesses may need guidance to apply the standard, Callaghan Innovation’s Digital and AI Whare Head, Sarah Sun, commissioned Craig Pattison to develop a series of fact sheets designed to break down key concepts. These resources provide practical advice and are freely available on Callaghan Innovation’s website. ‘We wanted to make them easy to read and accessible to encourage businesses to start applying the standard. The fact sheets support a call to action for businesses to adopt the standard and share user cases, helping others confidently integrate AI into their projects,’ says Craig.
Callaghan Innovation AI ISO standards (external link)
‘The project to develop the standard started in 2017 with the formation of the international committee. AI was in its relative infancy and has since rapidly developed. We wanted the fact sheets to support a call to action for people to start applying the standard and implementing the principles they contain. Being something new we want user cases so others can learn and feel more confident including AI in development projects.’
Agreed global guidelines
The ISO/IEC 42001 standard is especially valuable for businesses that operate globally, helping them adhere to evolving regulations such as the European Artificial Intelligence Act. This regulation categorises AI applications into unacceptable, high-risk, and lower-risk classifications.
‘Following the standard helps businesses meet these requirements and, importantly, provides evidence of compliance. Having an agreed set of guidelines accelerates AI adoption and scalability, whether using large language models developed in-house or publicly available tools like ChatGPT,’ explains Craig.
The standard was developed through six joint working groups across IEC and ISO, addressing critical areas like trustworthiness, ethics, sustainability, health informatics, safety, data processing, and conformity assessment. Additional working groups focused on foundational concepts such as risk management, AI frameworks, and computational approaches to ensure the standard's broad applicability across industries.
Raising the indigenous voice
‘One key contribution we gave was around the ethical considerations for indigenous language and culture, particularly New Zealand’s integral need for te ao and te reo Māori. Much of the international contributions had not considered indigenous aspects of AI. A lot of indigenous information has not been codified. Think of AI as a child learning a language. If it only ever learns English, how will it know what to do with requests in other languages.
‘With around 400 million people identifying as indigenous and being spread across 70 countries worldwide for AI to be accessible, trustworthy and fair it needs to avoid unconscious bias. For example where AI might be used for screening job applicants, without indigenous consideration, those using te reo might be left out of a hiring process. Or te reo speakers might not be understood by chatbots.
‘So, representing indigenous needs in AI is essential not only for New Zealanders but also for preserving and empowering indigenous and minority languages and cultures worldwide. This includes languages such as Louisiana Creole, the Inuit languages of Canada and Alaska, and Bahasa in Indonesia. Even widely spoken languages like Hindi and Mandarin must be considered to ensure AI is developed with global equity, future-proofing access for nations yet to fully integrate AI technologies. Without this consideration, AI risks perpetuating systemic exclusions in its learning models and outcomes.
‘Indigenous knowledge is structured differently from Western frameworks. Western knowledge systems often rely on large language models focused heavily on logic, abstraction, and reasoning, sometimes at the expense of capturing emotional context, spiritual values, and lived experiences. Consequently, AI can fail to interpret critical cultural nuances, thus reinforcing cultural biases and overlooking essential aspects of indigenous interpretation and knowledge systems.’
Supporting business innovation
The AI Blueprint for Aotearoa 2024 outlines sector-specific opportunities for AI investment, with a focus on agriculture, construction, and education. Federated Farmers, for example, highlights the diversity of farm ownership models in New Zealand—ranging from Iwi-based collective stewardship to family-run operations and corporate farms. AI systems designed for agriculture must account for these differences to avoid generic solutions that fail to reflect real-world complexities.
Governance frameworks should ensure AI systems are adaptable and capable of addressing both collective and individual perspectives. This requires careful consideration of sustainability, cultural stewardship, and short-term economic goals.
A critical area of AI governance also involves protecting indigenous cultural heritage. AI applications—such as those generating artwork—must embed protocols to respect privacy, cultural boundaries, and intellectual property. This includes compliance with local laws like New Zealand’s Privacy Act 2020 and Copyright Act 1994, alongside international standards under ISO/IEC 42001.
By fostering transparency and cultural awareness, AI governance frameworks can promote innovation without compromising indigenous values or intellectual property rights.
Written to work across sectors
Unlike many other standards, ISO/IEC 42001 crosses multiple industries. The AI Forum’s recent AI Blueprint for Aotearoa 2024 highlights current investments in AI and provides strategic guidance for businesses over the next five years. ‘There is already strong interest in sectors like agriculture, health, and education, with others wanting to move forward. This standard provides confidence and direction to implement AI policies and stay ahead,’ explains Craig.
Businesses can begin their AI journey by using the fact sheets developed through Callaghan Innovation to understand and implement AI management systems. ‘Together, we’re on a mission to help Kiwi businesses see that the standard is not a handbrake but an accelerator for their participation in the global AI economy,’ adds Craig.
The standard is currently under review for adoption as an AS/NZS standard, ensuring it meets the needs of Trans-Tasman industries. In the meantime, businesses can obtain the ISO/IEC version through Standards New Zealand.
Craig also encourages those working in the AI field to get involved with the AI Forum and the New Zealand mirror committee for Artificial Intelligence. ‘If you want to help shape the next iteration of the standard, we’d love to hear from you.’
Contact Craig Pattison craig.pattison@capabilitycollective.co.nz for more information.