Australia’s federal government wants AI developers to prove their safety systems work. It’s borrowing from the rulebooks that govern banks and aircraft. National standards for frontier artificial intelligence are due by the end of the year, with legislation to follow in 2027. Canberra says voluntary safeguards simply aren’t up to the job.
Assistant Minister for Science, Technology and the Digital Economy Andrew Charlton made the case in a speech in Sydney. He said Australia should put the “burden” where the risk is created. “AI needs regulating because its harms are severe, hard to undo, borne by people who never chose them and sometimes invisible until they arrive,” he said.
The Policy
The proposed approach doesn’t ban specific dangerous behaviours. Instead, companies would have to keep testing for risks and show they can prevent harm. Charlton said detailed prescriptive rules could be “out of date before the ink is dry,” a striking admission from a minister about the limits of rule-writing in a fast-moving field.
He pointed to systems-based regulation already used in workplace health and safety, supervision of banks and aspects of aviation safety. “The goal is to ensure companies are meeting the expectations of safety, without trying to specify every hazard,” he said. “The question isn’t only ‘did something go wrong’ but is there a serious system in place to detect risks and prevent incidents occurring.”
The model puts a compliance duty on firms and leaves the engineering to them. It also leaves a question unanswered: who will certify that a company’s systems are serious enough, and at what cost to developers?
What Prompted It
The push followed revelations that OpenAI agents gained unauthorised access to Australian government websites, including a Medicare statistics portal, during testing. OpenAI apologised to the government. The company has since backed mandatory safety requirements, independent assessments and incident reporting, while also calling for international standards.
Chief executive Sam Altman appeared at the United Nations last month and said the “most important decisions cannot be made by labs in San Francisco alone.” He said: “We need accurate and speedy incident reporting, classification and reporting protocols, so the world can learn from failures before they become catastrophes.”
Charlton argued the market can’t solve this by itself. “The market will not fix this alone, because the incentives reward speed and capability, and even the people at the top of the industry cannot slow down by themselves,” he said. “You cannot ask firms to volunteer against their own competitive interest amid a manic race involving trillions of dollars and then be surprised when they do not.”
Security and Sovereignty
Charlton also conceded the limits of regulation. Safety standards alone won’t stop “hostile states or criminals who were never going to comply.” “Relying on regulation alone is like building a fence around a horse, only to realise that horse has wings,” he said. He added that Australia would need to “host and build” AI technologies to influence how people develop and govern them. “This is our chance to shape this technology, for our standards to set the rules of the road globally over decades,” he said.
Shadow Foreign Minister Ted O’Brien pressed a more security-first line. Speaking at the University of Technology on Wednesday, he called for a government-to-government deal with the United States guaranteeing continued access to American technology, alongside the development of Australia’s own sovereign AI capabilities. “We can use AI as a shield to counter AI being used as a sword,” he said.
Charlton said Australia’s task wasn’t to “win” the race or “sit it out” but to “change it so that this technology is built to serve people.”
Why This Matters:
Canberra is preparing to impose a new compliance regime on one of the world’s most capital-intensive industries, and the details will decide whether Australia attracts AI builders or drives them elsewhere. A flexible, outcomes-based model beats a thick rulebook that dates quickly, and Charlton himself acknowledged that risk. But a duty to “prove” safety raises real questions about who audits, who pays and how much discretion regulators will hold. Charlton’s admission that regulation can’t stop hostile actors strengthens the case for O’Brien’s focus on secure access to allied technology and domestic capability. Businesses and investors should watch the end-of-year standards closely, because legislation in 2027 will lock in the costs.