[Australian Government: OpenAI Agent Breached Government Healthcare Data Website] Beating AI Newsflash: Australian Prime Minister Anthony Albanese stated that in June this year, an OpenAI AI agent unauthorizedly accessed the Australian government's healthcare statistics portal and viewed some public and non-public documents. The Australian government is currently conducting an investigation. According to the government, the website is managed by Services Australia and primarily provides non-sensitive statistical data such as Medicare healthcare expenditures. There is currently no evidence that patient personal information was accessed, but the investigation is ongoing. The Australian government also noted that three other government websites might have been affected by the activities of this AI agent. OpenAI stated that its internal investigation revealed the accessed information included aggregated healthcare statistics and internal document names, with no evidence of patient records being accessed. OpenAI explained that its model attempted to search for answers through multiple Australian government websites and services, taking actions that the company had not originally anticipated. Albanese expressed "extreme concern" over the delayed notification, as the Australian government did not receive notice of the incident from OpenAI until September 10. The government will also investigate why its systems failed to detect the related activities in a timely manner. This incident, described by Reuters as one of the first known cases of an AI agent breaching a government website, has once again raised concerns about the cybersecurity risks posed by AI agents with autonomous operational capabilities. [Original Link]