OpenAI agent hacked gov health portal but tech firm took months to tell officials
An OpenAI agent accessed public and non-public data on Australia’s Medicare portal during a health research task in June, officials have said.
Australian prime minister Anthony Albanese said the tech firm did not alert the Australian government to the breach until 10 September, when the company sent an email to a generic public mailbox.
He said he had told Altman he was disappointed it had taken the company “way too long” to inform the government of what had happened.
Deputy prime minister Richard Marles said the agent had also approached three other Australian websites and interacted with them in an authorised way, “in a way that a member of the public might”.
The autonomous AI agent was researching health and medical statistics when it accessed information on the Medicare portal run by Services Australia.
But when the agent sought information from the Medicare portal and was refused, Marles said, “it effectively hacked into that medical portal and got that information anyway”.
“It’s that unauthorised access which we are very concerned about,” Marles said.
“The impact is relatively minor, but the incident is very serious.”
“This situation is obviously unacceptable,” Albanese said.
“Today, I spoke with the CEO of OpenAI, Sam Altman, to express Australia’s extreme concern about this incident and I also expressed my disappointment that it took the company way too long to inform the government what had occurred.”
OpenAI said it was still investigating the incident and had found no evidence that patient records had been accessed.
The company said its review had identified activity involving several Australian government websites and services as its “models attempted to look up answers”.
Albanese said there was no evidence of a wider compromise of the government’s services network and that no personal information was believed to have been accessed, although investigations were continuing.
The disclosure came less than a day after Albanese joined 21 other countries, including Canada, Spain and Germany, in calling for “urgent global guardrails” on frontier AI models on the sidelines of the UN General Assembly in New York.
