For most of the past two years, Australia’s public conversation about artificial intelligence has swung between two poles. One camp treats the technology as an existential threat that needs tight rules before it does real damage. The other treats it as a once-in-a-generation productivity windfall that regulators should not slow down. A new analysis from the Lowy Institute argues that Anthony Albanese has finally nudged the country toward the more useful question, which is not whether to embrace AI but how to make sure Australia captures the gains rather than simply importing the risks.
The Lowy Institute piece lands at a moment when the government has been sending clear signals that AI sits near the centre of its economic agenda. The Prime Minister has tied the technology to the productivity story he wants to tell heading into the back half of the term, and Treasurer Jim Chalmers has made lifting Australia’s stubbornly flat productivity growth a defining goal. The Productivity Commission has already flagged AI as one of the biggest levers available, and the government has resisted pressure to rush a standalone AI Act of the kind the European Union has adopted.
What the analysis is actually saying
The core of the Lowy argument is that framing matters. By talking about AI in terms of national capability, jobs and economic weight rather than only in terms of harm, the Prime Minister has moved the debate onto ground where Australia can actually do something. A small open economy cannot set the global rules for frontier models the way Washington, Brussels or Beijing can. What it can influence is how quickly and how well its own businesses, hospitals, farms and public agencies put the technology to work, and whether any of the underlying capability sits onshore.
That is where the analysis turns from praise to warning. Opening the right debate is not the same as winning it. The piece cautions that Australia has a long habit of producing strategies, reviews and roundtables while the practical adoption of new technology lags well behind comparable economies. On that reading, the danger is not that the country regulates AI too harshly or too loosely, but that it spends the next few years discussing AI far more than it deploys it, and wakes up dependent on capability built entirely offshore.
Two ways to read the same moment
Not everyone reads the government’s posture so charitably. Safety-focused researchers and civil society groups have spent months arguing that Australia’s light-touch instinct leaves real gaps, particularly around automated decision-making in government, workplace surveillance and the use of AI in high-stakes settings like health and policing. From that vantage point, celebrating a shift toward the economic framing risks quietly sidelining the harms that motivated the debate in the first place. They point out that the country still lacks the kind of enforceable, AI-specific protections that a growing number of jurisdictions have put in place.
Industry sits on the other side of the same argument. Technology firms and many economists worry that a heavy compliance regime would land hardest on exactly the local start-ups and mid-sized companies that Australia needs to grow, while doing little to constrain the multinationals that already dominate the market. Their preferred path is to lean on existing consumer, privacy and anti-discrimination law, sharpen it where AI exposes weaknesses, and put the bulk of public effort into skills, infrastructure and adoption incentives. The Lowy analysis broadly aligns with the view that capability, not caution, is the scarcer resource for a country in Australia’s position, without dismissing the safety case outright.
What it means for Australia
The stakes here are concrete, not abstract. Australia’s economy is heavily weighted toward sectors where AI could deliver outsized gains, including mining, agriculture, healthcare, financial services and the sprawling public sector. Yet the country buys most of its frontier models, cloud capacity and specialist talent from overseas, which means the productivity upside can flow out as much as it flows in. The debate the Prime Minister has opened is really a debate about how much of the value chain Australia wants to hold, from data centres and energy through to sovereign models trained on local data.
That question connects to arguments already playing out across the sector. Researchers at the CSIRO have made the case that Australia’s best shot is not to chase the biggest general-purpose models but to build smaller, sharper systems tuned to national strengths. Infrastructure operators are pouring money into local data-centre capacity, while energy planners warn that the power demands of AI could collide with the country’s climate commitments if the build-out is not managed carefully. The Lowy piece effectively asks whether the political conversation is mature enough to weigh all of those trade-offs at once, rather than treating AI as a single yes-or-no decision.
There is also a sovereignty dimension that resonates strongly in Canberra. As allied governments tighten export controls and negotiate access to advanced chips, Australia’s ability to shape its own AI future depends partly on diplomacy and partly on domestic investment. Getting the framing right, as the analysis argues the Prime Minister now has, is the precondition for the harder budget and industry decisions that follow.
What’s next
The immediate test will be whether the rhetoric translates into policy detail. The government has signalled it favours adapting existing laws over a sweeping new statute, and the exact shape of any mandatory guardrails for high-risk uses is still being worked through. Expect continued pressure from safety advocates for clearer rules on automated decisions in government, and continued lobbying from industry for adoption incentives, skills funding and faster approvals for the data centres and energy projects the sector needs.
For all its praise of the framing, the Lowy analysis leaves the reader with a challenge rather than a verdict. Australia has opened the right conversation. Whether it can move from talking about AI to genuinely building and using it, at the scale a mid-tier economy needs to stay relevant, is the question the next couple of years will answer. On current form, the country is very good at the first part and has yet to prove it can deliver the second.
Sources: Lowy Institute (via GNews).
