For most of the past two years, the story of Australian artificial intelligence policy has been one of caution. Governments talked about guardrails, ran consultations on high-risk uses, and largely left the heavy lifting of adoption to the private sector. A new analysis published by SBS argues that the mood has changed, and that Canberra is now steering towards something more deliberate: a coordinated national push to build, adopt and govern the technology rather than simply react to it.
That shift has been building for months. The Albanese government has moved to set up a central Office of AI to pull together a policy effort that had been scattered across departments, and it has leaned harder into the language of productivity, sovereignty and national capability. The framing is no longer only about what could go wrong. It is increasingly about what Australia stands to lose if it sits on the sidelines while the United States, China and parts of Europe pour capital into models, chips and data centres.
What has actually changed
The clearest signal is one of posture. Where earlier statements emphasised risk and restraint, the current direction pairs adoption with ambition. Ministers now talk openly about lifting sluggish productivity growth, about keeping research and manufacturing onshore, and about making sure Australian firms are users and builders of AI rather than passive customers of overseas platforms. The rhetorical distance from the more hesitant posture of 2024 is significant.
Andrew Charlton, the assistant minister who has carried much of the AI brief, has repeatedly pressed the case that Australia needs to be more ambitious rather than merely safe, a theme that has run through recent government forums. Prime Minister Anthony Albanese has echoed the pivot in speeches that tie AI to jobs, investment and the country’s economic future. The message, in short, is that the government wants to be seen setting the pace, not applying the brakes.
Yet the SBS analysis is careful not to declare the job done. Its central point is that a change in tone is not the same as a change in outcomes, and that the hard questions sitting underneath the new strategy remain stubbornly open.
The questions that will not go away
The first is regulation. Australia still does not have a dedicated AI law. The government has flagged mandatory guardrails for high-risk uses and has canvassed everything from voluntary standards to targeted amendments of existing laws, but the final shape is unresolved. Businesses say they need certainty before they commit serious money, while civil society groups warn that a light framework could leave consumers exposed to automated decisions they cannot see or challenge. Splitting the difference has proven politically awkward, particularly on contentious fronts such as copyright and the use of Australian creative work to train foreign models.
The second is capability. This is where the research sector, including the universities and the CSIRO, enters the story. Australia has genuine strengths in AI research, from machine learning to applied work in agriculture, health screening and environmental monitoring. But researchers have long warned that funding is thin, that talent is being drawn offshore by better-resourced labs, and that the country lacks the sovereign compute to train large models at home. A national strategy that promises capability without addressing the plumbing behind it risks becoming an aspiration rather than a plan.
Two ways of reading the pivot
Supporters of the new direction argue that coordination is overdue. For years the criticism of Australian policy was that it was fragmented, slow and reactive, with different agencies pursuing different pieces of the puzzle. A central office and a clearer national framework, on this view, are exactly what a mid-sized economy needs to punch above its weight, concentrate scarce resources, and give industry the confidence to invest. They point to the data centre boom, the growing interest from global AI firms in Australian infrastructure, and rising enterprise adoption as evidence that the settings are finally aligning.
Sceptics counter that announcements have outpaced delivery. They note that a change in messaging costs nothing, and that the real tests are budgetary and legislative. Without sustained research funding, without a resolution on regulation, and without a credible answer on sovereign compute and energy, the danger is that Australia ends up hosting other countries’ AI rather than building its own. That distinction, between being a genuine builder and a well-serviced customer, has become the central fault line in the domestic debate.
Why it matters for Australia
The stakes here are not abstract. AI is already reshaping how banks handle customer service, how supermarkets run their supply chains, how farmers manage paddocks and how hospitals triage screening. The technology also sits at the centre of a national infrastructure question, because the data centres that power it are hungry for electricity and water at a time when the grid is already under strain. Every choice the government makes about regulation, funding and sovereignty flows through to jobs, prices, privacy and the country’s long-run productivity.
There is also a sovereignty dimension that resonates strongly in Canberra. Much of the compute, the frontier models and the capital behind the AI boom sits offshore. If Australia wants a say in how these systems are trained, deployed and governed, it needs domestic capability to bargain with, not just goodwill. That is why the research sector’s warnings about funding and talent carry weight well beyond the universities themselves. Capability is leverage.
What happens next
The next phase will be measured less by speeches and more by concrete decisions. Watch for the detail of any AI regulation, including whether high-risk uses attract genuine obligations or lighter-touch guidance. Watch the budget for whether the government backs its capability rhetoric with money for research, skills and sovereign compute. And watch how the copyright fight lands, because it will signal whether Canberra is willing to face down large technology companies to protect Australian creators and data.
The SBS analysis lands on a fair conclusion. Australia has, at last, chosen a direction. Whether that direction amounts to a real change of course, or simply a change of language, will depend on the answers to questions that are still very much unanswered.
Sources: SBS.
