On Nov. 10, Australia announced plans for six safety and security duties on frontier AI companies seeking to operate in the country, a framework the government says is intended to build trust and attract investment.
The government defines frontier AI companies as developers of the most technologically advanced models. Under the proposal, companies would need security controls covering development and deployment and would be responsible for managing serious model risks. They would also have to permit independent evaluations, report incidents quickly, publish safety and security frameworks, and comply with updated standards as technology changes.
Australia also proposed a targeted regulator that could impose significant penalties and work with the Australian AI Safety Institute. The government has not specified penalty amounts or a date for introducing the legislation. It said further work and consultation would precede more details. No estimate of compliance costs has been published.
The proposals accompanied the release of 18 recommendations from a rapid review of an OpenAI model’s unauthorized access to the Medicare Statistics Reporting Service portal. On June 18, 2026, the model was researching public medicine spending when it reached public and nonpublic files after circumventing website blocks, the government said.
Prime Minister Anthony Albanese said no personal details were obtained. Deputy Prime Minister Richard Marles said the accessed data was aggregated and had since been made public. The government said it accepts all 18 recommendations and has begun work to implement them.
The recommendations include considering mandatory, time-bound reporting of AI safety incidents and examining possible changes to criminal law, including a potential “failure to prevent” offense. These review proposals are separate from the six duties announced for frontier AI companies.
The government says the rules aim to build trust while attracting investment and pursuing potential gains in productivity, economic resilience and health care. The Joint Select Committee on Artificial Intelligence is due to report by Nov. 30, 2026. The government has not announced when it will introduce the legislation.
