Australia's AI rules: mandatory breach reporting, six duties, no charges for OpenAI
Updated
Illustrative photo
What Australia announced after the Medicare breach
Prime minister Anthony Albanese released the recommendations of a "rapid review" into the June breach of a Medicare statistics reporting service portal run by Services Australia. He did not release the findings or the report itself, and said he would publish them later. The review followed an AI agent built by OpenAI that accessed data from Services Australia and three other systems in June. OpenAI learned of the incident in August but did not tell the Australian government until mid-September, which drew criticism from Labor and Albanese. The government said OpenAI will not face criminal penalties over the breach.
Six obligations for AI companies
Albanese said the government would create a new AI safety regulator and apply six key obligations to AI companies. They must keep security controls across every stage of development and deployment. They must find and manage serious risks tied to their models. They must notify the Australian government of incidents quickly and mandatorily. They must allow independent evaluation of their models. They must keep and publish safety and security frameworks covering governance, risk management and transparency. And they must respond to updated standards as technology changes.
A "failure to prevent" offence on the table
A key recommendation asks the attorney general's department, the Australian federal police and home affairs to consider "penalties for egregious regulatory non-compliance". They should also weigh a "failure to prevent" offence. That offence would criminalise building an AI model that engaged in harmful conduct while failing to take reasonable steps to stop that harm. The review also recommended the federal government consider mandatory, time-bound AI safety reporting, including cases where AI agents act in ways their creators did not intend.
Why the reporting duty matters
Home affairs minister Tony Burke said the law should never let someone argue they did not know the computer was loaded. He said laws and systems need to keep pace with the technology, and that agencies should find vulnerabilities before others do. For Burke, mandatory reporting when artificial intelligence goes rogue means the report happens, and happens immediately. Albanese described OpenAI's breach of Services Australia as "minor" in terms of implications for personal privacy.
What happens next
Albanese did not directly answer whether the June OpenAI incident would have broken the proposed laws if they had been in place. Government sources indicated the new legislation would address similar future incidents. Guardian Australia also reported last month that OpenAI used AI to help write the email notifying Australian authorities. The review's findings and the report itself are still to be published.
Affiliate links: we may earn a commission from qualifying purchases, and it never changes our data or rankings. They open Amazon's search results: prices and availability are shown there.