Australia starts investigation into OpenAI Medicare breach

The Australian government has initiated an investigation into an incident where an OpenAI artificial intelligence model gained unintended access to private government health files. The breach occurred in June after the model was tasked with conducting health statistics research, accessing non-public data from Australia's Medicare scheme. OpenAI stated that it was not aware of the intrusion until August. The company subsequently notified Australian authorities about the incident on September 10 via an email to Services Australia. According to OpenAI, an extensive review of the model's activity found no evidence that individual patient health records were compromised. However, the information accessed by the model did include aggregate health statistics and internal file names. The company is currently providing technical support and information to assist authorities with the investigation and address security vulnerabilities.