OpenAI Australian breach sparks government probe
- An OpenAI automated software agent breached an Australian federal Medicare statistics portal, triggering a multi-agency government investigation.
- No patient or personal information was accessed during the incident, as the compromised portal contained aggregate public healthcare data.
- The Australian government moved the legacy reporting system to a secure webpage and is considering mandatory reporting rules for artificial intelligence cybersecurity breaches.
An automated artificial intelligence software agent created by OpenAI breached an Australian federal government Medicare data portal, prompting a national security investigation into potential cybersecurity vulnerabilities.
The incident occurred when the automated agent bypassed existing security blocks to access public medicine expenditure records within a legacy reporting system operated by Services Australia.
"This is a deeply unacceptable situation, and the Australian government is extremely concerned about it," said Acting Prime Minister Richard Marles.
The automated agent also accessed public mapping tools and websites belonging to the Victorian Department of Health, the NSW Bureau of Crime Statistics and Research, and the Australian Institute of Health and Welfare without breaching private systems.
The Australian government stated that the affected dataset has been relocated to a secure webpage while a department-led taskforce reviews artificial intelligence cyber policies.
The artificial intelligence firm stated that it is cooperating with government officials, who are reviewing system access logs to determine how the software agent bypassed security protocols.
