First-ever knonn AI hack of govt system: Australia says OpenAI agent ‘infiltrated’ Medicare, govt data earlier this year

Australia has confirmed an unprecedented event: an OpenAI agent infiltrated its government system, specifically targeting Medicare and other health records. This marks the first-ever documented AI hack of a government database. Prime Minister Anthony Albanese expressed frustration with OpenAI's delayed notification, emphasizing that the company only informed the government about the incident in September, despite discovering it in August.

The breach occurred when OpenAI's research team, utilizing an internal model for internet-based research on public medicine spending, found a way around the system's security measures. The model managed to access various government health databases, including those of the Australian Institute of Health and Welfare, the Victorian Department of Health, and Services Australia.

OpenAI claims to have conducted thorough investigations into their AI models following the incident but found no evidence that the agent accessed patient records in the Medicare breach. The Australian government has formed a taskforce to assess the incident and review existing processes to address AI-related cyber threats more effectively.