Australia has spent the better part of two years talking about how to rein in artificial intelligence without deciding what, if anything, to write into law. That may be about to change. Prime Minister Anthony Albanese has signalled that his government intends to enact legislation to govern AI, a move that would shift the country from voluntary codes and consultation papers towards binding rules for the first time.
The comments, reported by Yahoo, land at a moment when Canberra has been under pressure from two directions at once. Businesses want certainty about what they can build and deploy, while unions, creators and digital rights advocates have grown impatient with a policy process that has produced plenty of discussion papers but few enforceable obligations.
How we got here
The federal government began seriously grappling with AI regulation in 2023, when it opened a consultation on safe and responsible use of the technology. That work produced an interim response acknowledging that existing laws, covering privacy, consumer protection, corporations and copyright, were not written with generative AI in mind and left gaps. From there Canberra floated the idea of mandatory guardrails for AI used in high-risk settings, alongside a voluntary safety standard that companies could adopt in the meantime.
The through-line of that approach has been caution. Rather than copying the European Union’s sweeping AI Act wholesale, officials have leaned towards a risk-based model that would apply the heaviest obligations to uses with the greatest potential for harm, think healthcare, employment decisions, policing and critical infrastructure, while leaving lower-stakes applications largely untouched. What has been missing is a firm commitment to actually legislate. The Prime Minister putting his own weight behind new laws changes the temperature of that debate.
The case for acting now
Anthony Albanese and his government have framed AI as both an economic opportunity and a source of risk that markets will not manage on their own. The productivity argument is central: AI is one of the few levers economists believe could lift Australia’s stubbornly flat productivity growth, and the government has repeatedly linked the technology to its broader economic reform agenda. But the same ministers have been forced to respond to a steady drumbeat of harms, from deepfakes targeting school communities to concerns about AI in recruitment, banking and medicine, all of which have featured prominently in Australian public debate over the past year.
Supporters of legislation argue that clear rules are good for business, not a brake on it. When companies know where the legal lines sit, they can invest with confidence rather than second-guessing whether a product will be pulled up after launch. Consumer and civil society groups make a complementary point: voluntary standards have no teeth, and a company that ignores them faces no consequence beyond reputational risk. Binding law, in their view, is the only way to guarantee that safeguards such as transparency, testing and human oversight actually happen.
The case for caution
Not everyone is convinced that fresh AI-specific legislation is the right instrument. Parts of the technology sector and the business community have warned against rules that could entrench incumbents, deter local start-ups and duplicate protections that already exist under privacy and consumer law. Their preferred path is to fix the gaps in existing frameworks and reserve new obligations for genuinely high-risk uses, rather than creating a broad new compliance regime that smaller firms would struggle to navigate.
There is also a timing problem that regulators everywhere are wrestling with. The technology is moving faster than any legislative cycle, and rules drafted around today’s chatbots and image generators risk being outdated by the time they pass Parliament. Critics on this side of the argument favour principles-based law that sets expectations without prescribing specific technologies, so the rules do not need rewriting every time a new model arrives. The counter-view, familiar from the debate over the EU’s approach, is that principles without detail leave too much to interpretation and enforcement becomes toothless in practice.
What it means for Australia
For Australian businesses, the prospect of binding AI law raises immediate questions about compliance, cost and competitiveness. The country is a heavy adopter of AI tools built overseas, which means much of what Australians use is designed for other regulatory environments. Any domestic law will have to reckon with that reality, deciding how far it can hold multinational developers to account for products built abroad and deployed here. Sectors that have moved fastest, including the major banks, insurers, health providers and government agencies experimenting with AI, would be the most exposed to a new regime, and they will be watching closely for detail on scope and timing.
There is a sovereignty dimension too. Australia has limited leverage over the global giants that dominate AI development, and going it alone with strict rules risks either being ignored or discouraging investment. Aligning with international frameworks, whether the EU’s model or emerging standards being developed elsewhere, could give Australian rules more force while easing the burden on companies that already comply overseas. Getting that balance right matters for a mid-sized economy that wants the productivity upside without becoming a rule-taker on terms set entirely offshore.
For workers and consumers, the stakes are more personal. The harms that have dominated headlines, automated decisions that affect jobs and credit, synthetic media used to deceive, and the use of personal data to train models, are precisely the areas where Australians have told successive surveys they want stronger protection. Whether the government’s legislation delivers meaningful safeguards or settles for a lighter touch will shape how much trust the public places in AI as it spreads deeper into everyday life.
What’s next
The Prime Minister’s signal is a statement of intent rather than a bill, and the hard work lies ahead. The government will need to settle the fundamental design choices it has been circling for two years: whether to legislate through a standalone AI Act or by amending existing laws, how to define high-risk uses, what obligations to impose on developers versus deployers, and which regulator would enforce the rules. Expect renewed consultation with industry, unions and civil society before any legislation reaches Parliament, and expect the familiar tension between speed and caution to play out at every step. After a long stretch of talking, the country now has a clearer sense of direction, even if the destination is still being drawn.
Sources: Yahoo via GNews.
