Australia will require AI companies to demonstrate they are properly testing models and preventing harm under proposed new laws modelled on banking and aviation safety rules, the federal government has signalled.
Assistant Minister for Science, Technology and the Digital Economy Andrew Charlton outlined the case for tougher frontier AI oversight in a speech in Sydney, arguing Australia should put the "burden" where the risk was 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," Mr Charlton said. "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."
Rather than trying to keep pace with the technology by banning specific dangerous behaviours, Australia's approach would require AI companies to continually test for risks and show they can prevent harm.
Mr Charlton said there was not enough incentive for companies to comply with "voluntary" regulations or codes when it came to frontier AI risks.
"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," he said.
He also warned detailed prescriptive rules could end up "out of date before the ink is dry" due to the pace of the technology's development.
"For these reasons, there is much that Australia can learn from systems-based regulatory approaches, that puts the onus and accountability on companies for managing how they develop products and services safely," he said.
Mr Charlton said this model was 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 push for tougher regulation has been sharpened by revelations that OpenAI agents gained unauthorised access to Australian government websites, including a Medicare statistics portal, during testing.
OpenAI has now backed mandatory safety requirements, independent assessments and incident reporting, the government has said.
The government is due to finalise the national standards by the end of the year.