Australia to make AI companies report rogue model behavior after OpenAI agent breached a government site

Prime Minister Anthony Albanese said Tuesday that Australia accepted all 18 recommendations of a rapid review into the June incident in which an OpenAI agent, researching public medicine spending, got around access blocks and pulled non-public aggregate data from a Services Australia Medicare statistics portal. The changes include mandatory reporting by AI companies of misaligned behavior and cyber incidents, independent evaluations of frontier models, and consideration of criminal-law reforms for serious misconduct. A new AI safety regulator with power to impose penalties will enforce the rules. OpenAI didn't notify the government until September.