For most of the past two years, the Australian conversation about artificial intelligence has been framed as a matter of catching up. The models were built elsewhere, the compute sits elsewhere, and the money that funds the whole enterprise flows through a handful of American and Chinese firms. So when Anthony Albanese argues that Australia can help write the rules rather than simply live under them, it is worth pausing on both the ambition and the gap between saying it and doing it.
That is the tension at the centre of a recent piece in The New Daily, which puts a blunt proposition to readers: the Prime Minister says Australia can set the terms for AI, and the open question is whether the country actually will. It is a useful reframing, because it moves the debate away from the familiar hand-wringing about whether Australia has missed the boat and towards something more actionable, namely what a middle power with strong institutions and a sophisticated economy can realistically shape.
From rule-taker to rule-maker
The case for optimism is not fanciful. Australia has form when it comes to punching above its weight on technology governance. The News Media Bargaining Code forced Google and Meta to the table over payments to publishers, and while the outcome was messy, it demonstrated that Canberra could impose terms on companies far larger than any Australian firm. On privacy, on financial regulation and on consumer protection, the country has a habit of building frameworks that others study.
Anthony Albanese has leaned into that history. His government has talked up sovereign compute, backed data-centre investment, and signalled that it wants Australian industry, health and education to be users and shapers of AI rather than passive consumers of foreign products. The logic is that a country with deep pools of skilled labour, world-class research institutions and a stable rule of law does not have to accept whatever Silicon Valley ships. It can insist on transparency, on safety testing, on local data handling and on rules that protect workers and consumers.
There is also a strategic argument. As democracies coordinate on AI safety through forums that grew out of the Bletchley process, Australia has a seat at tables where the norms are being negotiated. Being in the room matters, because the standards that harden into international practice over the next few years will be difficult to unwind later.
The case for scepticism
Against that sits a harder-nosed view, and it is one worth taking seriously. Setting terms requires leverage, and leverage in AI comes from compute, capital and talent, three things Australia has in modest supply relative to the giants. The country does not build frontier models. It does not manufacture the advanced chips that train them. Much of the cloud infrastructure that hosts Australian workloads is owned by American hyperscalers. Rhetoric about sovereignty runs into the reality that the underlying stack is largely rented.
Critics also point to a pattern of consultation without conclusion. The former industry minister, Ed Husic, ran an extensive process on safe and responsible AI, floated the idea of mandatory guardrails for high-risk uses, and canvassed whether Australia needed a dedicated AI Act. Yet the country still lacks binding, AI-specific legislation, relying instead on a patchwork of existing laws stretched to cover new problems. Business groups warn against over-regulation that could deter investment, while unions and civil society argue the vacuum leaves workers and citizens exposed. Setting terms means choosing, and choosing tends to annoy someone.
The Productivity Commission has added a further wrinkle by cautioning against rushing to a heavy new regulatory regime, arguing that poorly designed rules could smother the very productivity gains AI promises. That advice pulls in the opposite direction to those demanding a strong statutory framework, and the government is caught between the two.
What it means for Australia
For Australian businesses and workers, the stakes are immediate. Firms deploying AI want certainty about what is permitted, how liability is assigned and what standards they must meet, because ambiguity is its own kind of tax. Employers rolling out automation across banking, retail, logistics and the public service are making decisions now that will shape jobs for a decade, often ahead of any clear rulebook. The recent modelling suggesting AI could reshape a large share of Australian roles has only sharpened the sense that policy is trailing practice.
There is a sovereignty dimension too. Australian data, from health records to government services, increasingly passes through AI systems, and where that data lives and who can access it are not abstract questions. The push for local data centres and sovereign cloud capacity, backed by billions in recent infrastructure announcements, is partly about jobs and partly about not being wholly dependent on facilities and rules set overseas. Setting terms, in the practical sense, may look less like grand declarations and more like procurement standards, data-residency requirements and safety obligations baked into the contracts government and large enterprises sign.
The federation adds another layer. New South Wales has been striking its own arrangements with major AI providers, and states are moving at different speeds on adoption in schools, hospitals and transport. A national ambition to set terms sits uneasily with a system where much of the actual deployment happens at state level, which means coordination is as much a test of the idea as any single piece of legislation.
What comes next
The next twelve months will be revealing. The government has flagged a national capability plan and continued work on where, if anywhere, mandatory guardrails should apply. Watch for whether that work produces binding obligations or another round of principles and voluntary standards. Watch too for how Canberra handles the tension between the Productivity Commission’s caution and the appetite among some ministers and stakeholders for firmer rules.
The deeper point in The New Daily’s framing is that ambition and capability are not the same thing. Australia plainly can influence how AI is governed, through standards, procurement, alliances and the credibility of its institutions. Whether it will depends on choices that are political rather than technical: whether to legislate, whether to invest in genuinely sovereign infrastructure, and whether to accept the trade-offs that any real position on AI demands. Saying the country can set the terms is the easy part. The harder work is deciding what those terms should be and then holding the line when powerful interests push back.
Sources: The New Daily.
