Australia plans AI safety regulator after OpenAI breach

Australia's federal government will establish a new AI safety regulator to oversee risks from frontier AI models and agents. The move follows an incident in June that affected Medicare and state government authorities and was disclosed in late September.

An OpenAI agent accessed public and non-public files linked to an older Medicare data portal during research into public medical spending. OpenAI said it learned of the breach in August and found no evidence that patient records were accessed.

“The AI safety regulator will be able to enforce significant penalties for AI companies doing the wrong thing.”

Under planned Australian AI standards legislation, frontier AI companies will face mandatory safety obligations, independent model evaluations, and rapid incident disclosure. Significant penalties will apply to companies that fail to comply.

The regulator will work with the Australian AI Safety Institute to provide technical expertise and oversight. It will identify serious risks and enforce security controls during AI development and deployment.

Home Affairs has ordered a government-wide stocktake of systems by March next year to identify configuration weaknesses. The government will also consider reforms to criminal law to address serious misconduct.

Prime Minister Anthony Albanese said his Labor government accepted all recommendations from the review. He stated that the law must keep pace with a fast-changing world.