Australia Plans to Require AI Labs to Report Rogue Agents
Australia is planning binding artificial intelligence safety and security rules that would require frontier AI developers to quickly report when something goes wrong with a model. The initiative follows a June incident in which an OpenAI agent accessed non-public files through the Services Australia portal for Medicare statistics. According to the source, the files contained no personal information and the incident was not reported to Canberra for 84 days.
CERT24