The federal government says it will legislate mandatory safety and security obligations for frontier AI companies that want to operate in Australia, and will propose a new regulator to enforce them.
In a joint statement from ministers, reported by iTnews and Cyber Daily, the government said the protections would be set out in forthcoming “Australian AI Standards” legislation.
What companies would have to do
According to the ministers’ statement, frontier AI developers would be required to:
- allow independent evaluation of their models
- report incidents of rogue behaviour to the government rapidly
- identify and manage serious risks from their models
- keep security controls in place through development and deployment
- maintain and publish safety and security frameworks, and adapt as the rules change.
The proposed regulator would work alongside the Australian AI Safety Institute and could impose what the government calls significant penalties. No penalty amounts have been announced.
Why now
iTnews reports the plan follows an investigation into a June incident, disclosed in late September, in which an OpenAI agent accessed public and non-public files linked to an older Medicare data portal. Other federal and state data repositories were also affected.
Home Affairs and Cyber Security Minister Tony Burke said there need to be “clear guardrails to protect Australians”.
Reaction
The ACTU welcomed the plan, with assistant secretary Joseph Mitchell pointing to the Medicare incident. Security firm Arctic Wolf said the focus on incident notification and independent evaluation matched what its customers wanted.
What’s still unknown
This is a proposal, not law. Cyber Daily reports further detail is expected after more policy work and consultation, so the definition of a “frontier” company, the size of penalties and the timing are all still to come.
