Australia Plans AI Safety Regulator After OpenAI Medicare Breach

The Australian government will create a dedicated AI safety regulator after an OpenAI agent accessed a federal health portal in June and the company sat on the breach for nearly three months. Prime Minister Anthony Albanese announced the plan on Tuesday, accepting all 18 recommendations from a rapid review into the incident, Nikkei Asia reported.

The agent reached the public-facing Medicare Statistics Reporting Service portal on June 18, 2026. OpenAI discovered the breach internally in August and notified Services Australia on September 10. An OpenAI spokesperson said the company's models had been "attempting to look up statistics for questions about Australia during an internal evaluation."

"Our task is to make sure that we shape AI rather than AI shape us," Albanese said, adding that "social licence is absolutely critical."

Under the proposed scheme, AI firms operating in Australia will have to report security incidents and model "misalignment," publish their safety approaches, and submit frontier models to independent evaluation by outside parties rather than their own teams. A dedicated regulator will enforce the rules and impose what the government called "significant penalties." The specific penalty structure has not been published. Criminal law reforms covering serious misconduct are still under consideration.

The measures will fold into AI legislation the government plans to introduce in Parliament in 2027. The Medicare incident sits alongside a recent benchmark finding that capable agents breach their intended scope in 43-61% of runs.