The Australian Prime Minister has announced a new ‘Office of AI’ that will develop a single, national framework for the rollout of the technology.
The framework will bring together work from different ministries to address AI’s impact on climate change, energy, education, jobs, copyright, and defence.
There will be clear rules for large data centres in Australia: where they are built, and the power and water they use, the PM said.
The SMC asked NZ experts to comment.
Angus Dowell, PhD candidate in Economic Geography, University of Auckland, comments:
“The speech by Australian Prime Minister Anthony Albanese is highly significant. It moves beyond narrowly treating AI as a question of innovation or adoption, instead recognising its reach across labour, copyright, national security, energy, water and regional development and acknowledges the need to secure long-term value over a short term boom from construction in data centres.
“But most consequentially, the speech brings the politics of ownership, control, and public value to the centre of the AI infrastructure debate. He takes aim at one of the central issues plaguing the AI economy at large – big-tech, where the risk, he argues, is in potentially ‘subcontracting our sovereignty and security to the control of foreign monopolies’. His counterexample is the National Broadband Network, whose continued public ownership he presents as a means of retaining national control over foundational digital infrastructure even as communications networks become increasingly entangled with private and overseas interests.
“In this respect, the speech reads as an implicit rebuke to parts of Australia’s tech establishment who want otherwise. Where Tech Council of Australia chair Scott Farquhar earlier argued for deregulated ‘digital embassies’, faster approvals, and to loosen copyright law to attract foreign data-centre investment, Albanese offers a more conditional bargain in which he argues Australia should use its land, minerals, abundant solar, and political stability to secure better terms for local development.
“Of course, it remains to be seen what the terms of local development will actually be, and who gets access to shape them. These questions will determine whether the asymmetries of the AI economy can be meaningfully evened out on Australian soil, but for now the speech marks a significant and positive shift in the debate. New Zealand’s lesson is to avoid turning its budding AI economy into a race to the bottom for foreign investment, where cheaper power and faster consenting become the main selling points.”
Conflict of interest statement: Angus has declared that he has no conflicts of interest.
Dr Andrew Lensen, Senior Lecturer in Artificial Intelligence, School of Engineering and Computer Science, Victoria University of Wellington, comments:
“It is natural that we will see increased political attempts to control the AI agenda across the world, as the public’s concern about the impact of AI continues to grow. While the US, under Trump, has sought to promote the interests of AI companies, this move by the Australian PM seems more focused on protecting their citizens.
“We have seen Australia confront Big Tech before (e.g. taking action to protect news publishers from scraping) with some success, so it will be interesting to see how far they’re willing to stand up against the powerful AI companies — and how successful they are, as a reasonably small nation. There is a real resentment by the general public against the impact of AI, so I think we’ll see more and more left-wing governments take these sorts of policy positions.
“In contrast, we see little political leadership in New Zealand. None of our major political parties have standalone AI policy released for the 2026 election, despite AI having a major impact on key policy issues like the cost of living, jobs, the environment, and Māori sovereignty. We are continuing to fall behind on the opportunity to manage the risks of AI in New Zealand, with potential harms increasing by the day. For example, why are we allowing an AI data centre to be built in Southland that will drive up power prices, use drinking water, and export profits overseas for only a few local jobs? Australia is, at least, considering stricter requirements on AI data centres and requiring benefits to the communities that they will be built in.”
Conflict of interest statement: “Co-director of LensenMcGavin AI (AI consultancy).”
Professor Alistair Knott, Centre for Data Science and AI, Victoria University of Wellington, comments:
“New Zealand needs a government AI office – and can readily get one. For several years, there have been calls for the NZ government to establish a formal AI office of some kind. Some are calls for a body to regulate the government’s own uses of AI (see e.g. my group’s 2019 recommendation). Some are calls for the creation of an AI Safety Institute, to develop policy for the newer more powerful forms of AI (see e.g. this call). Some are more broader calls encompassing both of these (see the recent ‘Regulate AI‘ letter to party leaders).
“The government has responded with a ‘light touch’ approach to AI oversight, articulated in Judith Collins’ guidance document of June 2024 and the ‘AI Strategy‘ of June 2025. But their actions are moving in the direction of structured institutions. These are inevitable moves, given the rapid advance of AI technology, and its increasing pervasiveness in all aspects of commercial, public and private life.
“The government’s decision to centralise government digital delivery in a single agency in the Public Services Commission is particularly significant. A recent ‘reset plan‘ for this agency highlights the need for the need for a focus on AI in priority projects. (This plan explicitly looks to Australia’s Digital Transformation Agency as a model.) A national centre of AI expertise is also part of the Institute for Advanced Technology currently being set up by MBIE.
“Between them, these institutions provide much of the structure and expertise needed to create a national AI Office for New Zealand. The government should build on these existing institutional initiatives, to set one up.”
Conflict of interest statement: “I’m in one of the groups applying for MBIE’s AI Platform fund.”
Our colleagues at the AusSMC have also gathered comments. A small selection follows.
Professor Lisa Given, Director of the Centre for Human-AI Information Environments and Professor of Information Sciences at RMIT University, comments:
“The Australian government’s announcement of a new, central Office of AI, to sit in the Department of Prime Minister and Cabinet, is a welcome move that will provide clarity and guidance for businesses and communities. The Office will coordinate work across government departments and create new Australian standards for AI adoption.
“As technology companies increasingly look to build data centres in Australia, communities have expressed significant concerns about energy and water use, and placement near residential neighbourhoods. Australians are also worried about disruption from AI technologies, raising concerns about potential job losses, copyright infringements of material used to train AI models, and data privacy.
“Addressing such varied challenges, particularly when AI technologies are evolving at such a fast pace, warrants the coordinated approach this office will take. Given the known risks posed by generative AI technologies, for example, from deepfake images that mislead customers, to hallucinated content that misinforms chatbot users, Australians need government intervention to protect them from harm and ensure AI use is in their best interests.
“The creation of this office marks a significant shift in the government’s overall approach to governing AI, towards being more hands-on and proactive. In 2024, the government planned to regulate high-risk AI implementation but later abandoned that plan. The government also proposed – and then paused – digital duty of care legislation, to hold technology companies to account for harmful content and tool designs. By creating this office (and by saying digital duty of care legislation will be reintroduced this year) the government has recognised the need for coordination and intervention to address the varied risks posed by AI, while also supporting its potential benefits.”
Conflicts of interest: None declared.
Dr Eden Li, Senior Lecturer in the School of Business & Law at Edith Cowan University, comments:
“The establishment of an Australian Office of AI and a national AI framework represent a significant step towards Australia’s approach to AI future.
“Two things would be critical:
“1. It will be important to adopt a balanced view that recognises both the opportunities and the potential risks/cautions associated with AI. Its success should not be assessed through economic outcomes alone. Equal attention must be given to how AI affects people’s experience of work, communities, inclusion, and wellbeing.
“2. A collaborative and transparent approach will be critical. The national framework should be developed through meaningful engagement with industry, unions, community representatives, academics, and regional stakeholders. This will help ensure that Australia’s approach to AI is practical, inclusive, and equitable, to the ways in which AI may affect people, organisations, and communities.”
Conflicts of interest: None declared.
Dr David Tuffley, Adjunct Senior Lecturer in the School of Information and Communication Technology at Griffith University, comments:
“Establishing an Office of AI within Prime Minister and Cabinet is the right thing to do, but making it work will be easier said than done. Until now we have approached AI governance in a piecemeal fashion with departmental initiatives and voluntary guidelines. Having a unified national framework would give business and researchers the regulatory basis that they can proceed with. This would allow Australia to proactively shape the technology rather than simply using it as supplied.
“Declining to give AI firms a blanket waiver to use Australian data is a good move that will protect creators of digital assets. But the government will need to define what lawful access looks like and that’s not so easy. If our licensing pathways are difficult to use because they are unclear or convoluted then the developers will go offshore and Australia will be left to import the finished product but without any of the economic benefit.
“The test for the new Office of AI will be whether it can find the ‘sweet spot’ between the risks and the real cost of moving too slowly. Regulation that arrives late or lands heavily will not protect Australians. All it will do is make sure the technology that shapes our lives will be built elsewhere, under someone else’s rules.”
Conflicts of interest: None declared.
Dr Raffaele Fabio Ciriello, Senior Lecturer in the Discipline of Business Information Systems (BIS) at the University of Sydney Business School, comments:
“This announcement is too little, too late. The Australian government is finally waking up to the AI reality, but it continues to approach the challenge primarily as one of attracting investment with minimal risk mitigation.
“We have seen this model before in the gas industry. Foreign corporations use Australian resources, infrastructure, and labour, export much of the value, and leave Australians carrying substantial environmental and social costs. AI risks repeating and amplifying that mistake. Its data centres consume Australian land, water, and electricity, while its models draw on our research, creative work, data, and intellectual labour. Yet ownership, control, and profits remain concentrated in a handful of foreign corporations and their executives.
“The gas analogy, however, only goes so far. AI is not merely another extractive industry. It increasingly shapes how Australians work, learn, communicate, relate, and access or produce vital information. Dependence on privately controlled AI infrastructure is therefore a question of democratic and national sovereignty, not only taxation or regulation.
“Rejecting a copyright free-for-all is necessary, but far from sufficient. Australia needs collective sovereignty over AI infrastructure through meaningful public and community ownership, genuine citizen control, and a fair distribution of benefits according to social need. Otherwise, we will once again socialise the costs while privatising the gains.”
Conflicts of interest: None declared.
Professor Joel Pearson, Deputy Director of the AI Institute and researcher in human readiness and neuroscience, The University of New South Wales, comments:
“The Office of AI is the right instrument. Putting it in Prime Minister and Cabinet is the right call, because AI is not an industry portfolio, it is a whole-of-society one. The question now is what sits inside it. Standards design covers the technology. Nothing yet covers the people. An Office of AI without a human transition function is an office for managing machines in a country made of humans. Australia now has a body responsible for coordinating how AI is built here. We still don’t have one responsible for how Australians get through it.
“The Prime Minister is right that the world should have acted on social media a decade earlier. But look closely at what we actually got wrong. We didn’t get the infrastructure wrong. We got the psychology wrong. The harms were developmental, cognitive and social, and they arrived years before we had the evidence or the institutions to answer them. If we take that lesson seriously, we don’t just build better rules for data centres. We build the capability to see human harm coming before it shows up in a statistic. The social media reckoning was not a failure of engineering standards. It was a failure to understand minds. We are about to make the same category error at a much larger scale.
“Today we learned where data centres can be built, what power they must generate, how much water they may use, and who owns a song. Every one of those is a rule about the technology. Not one is a plan for the people the technology is about to move through. We are regulating the footprint and ignoring the population standing in it.
“The Prime Minister noted that extremists and state actors already use AI to target young people with propaganda. Every dollar of that threat lands in a human mind. You cannot firewall a population. The defence against synthetic persuasion is a citizenry equipped to recognise it, and that is a psychological capability, not a technical one. It is currently nobody’s job to build it.
“We want AI to support and create good jobs, not replace them.’ Everyone wants that. Wanting is not a mechanism. Data centres now have a legal obligation to put more power into the grid than they take out. That’s a mechanism. There is no equivalent obligation, anywhere in this framework, to give a single Australian worker warning, retraining, or support before their role is automated.”
Conflicts of interest: None declared.