Anthony Albanese says an artificial intelligence agent developed by OpenAI hacked Medicare in June and the tech giant notified the government earlier this month using an email sent to a public-facing mailbox.
Australia’s prime minister made the comments at the UN summit in New York, saying it appeared no personal information had been accessed in the AI breach.
Investigations assisted by the Australian Signals Directorate were ongoing, however, into how the agent had infiltrated Australia’s government-run universal health care system.
“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.”
Albanese said both the delay and the manner in which OpenAI notified the government was disappointing.
“It took until ۱۰ September before there was any notification at all. And the notification was an email sent to just the public mailbox,” Albanese said.
Services Australia then reported the notification to the Australian Cyber Security Centre on ۱۵ September.
A taskforce has been launched to explore the “legal situation” of the breach.
How the hack happened
Albanese said the OpenAI agent had gained unauthorised access into the public-facing Medicare statistics reporting service portal, which is administered by Services Australia.
The portal contains non-sensitive Medicare information relating to data and statistics, such as spending.
Albanese said the agent had accessed “public and non-public files.”
“A forensic investigation, aided by the Australian Signals Directorate, is now under way to ascertain more information, including what other government systems were affected,” Albanese said.
He said there appeared to be no broader compromise to the Services Australia network.
The deputy prime minister, Richard Marles, said the government learned of the June breach “a couple of weeks ago” and ministers were informed last week.
Marles said the agent was non-human and called the breach a “very serious incident”.
“What we have seen is unauthorised access in to an Australian government website and that is completely unacceptable and we have made that clear to OpenAI,” he said.
The AI agent had been given “a benign task” of researching health and medical statistics, and in doing so had approached three other Australian websites as well as the Medicare portal: the Victorian Department of Health, the New South Wales government website, the Australian Institute of Health and Welfare, Marles told ABC RN on Thursday morning.
“In relation to those three, it interacted in a way that a member of the public might, so it only acted in an authorised way. But in relation to the Medical Portal of Services Australia, it sought information, information was not given, and then it effectively hacked into that medical portal and got that information anyway,” Marles said.
“It’s that unauthorised access which we are very concerned about. … The impact is relatively minor, but the incident is very serious.”
‘Legal situation’ under investigation
Marles said the government had established a taskforce that will be led by the the Department of Prime Minister and Cabinet and working with the Australian Signals Directorate and the AI Safety Institute.
“We will look at what is the legal situation in respect of this and what it means to have gained an unauthorised access, albeit in an unintended way.
after newsletter promotion
“This is a warning about the fact that artificial intelligence, which is a technology that has huge opportunity to benefit humanity, has to be developed with enormous care. There have to be guardrails and safety measures in place which are way ahead of the capability which is being developed,” Marles said.
In a statement, OpenAI spokesperson Drew Pusateri said the company was conducting an extensive review of “misaligned model activity during training and evaluation” and was “notifying third parties when our review identifies potential impacts to their systems.”
“During this review, we identified activity involving several Australian government websites and services as our models attempted to look up answers, and available statistics for questions about Australia during an internal evaluation,” Pusateri said.
“In the course of that, our models took actions we did not intend.”
He said the review had found no evidence of patient records being accessed.
“The information accessed included aggregate health statistics and internal file names.”
Pusateri said OpenAI was supporting investigations and that its review was ongoing, adding it was committed “sharing what we learn as that work continues”.
Australians should be ‘worried’
Ed Santo, former human rights commissioner and co-founder of the Human Technology Institute, said Australians ought to “be worried”.
“I think we’ve got some serious problems,” he told ABC on Thursday morning.
“This is the first time that this has happened in any major way to an Australian database, and it’s one of the most sensitive databases that we have in government.”
OpenAI had “instructed their AI agents to go looking for information about the public health system in Australia. When the AI agents couldn’t easily find that information, they decided to break the law,” said Santo.
Describing this behaviour as “misaligned model activity” was using “a very euphemistic term”, said Santo.
“That’s ‘misaligned’ in the sense that that was not the intention of Open AI, that it would break into the database. That’s the first problem. The second problem is that these AI agents are hugely capable of hacking.”
The Australian government needed to ensure that the relevant laws and enforcement of them were “up to date and are capable of addressing this kind of unprecedented cybersecurity threat”, Santo said.
“If we were in the bricks and mortar world, if an employee of a company went and broke the law, particularly if it broke the criminal law, then that individual would suffer serious legal consequences, but so would their company. And so we need to make sure that that is true also in this era of AI,” Santo said.
“So if a company is going to rely on AI agents to do all kinds of things that carry risks, then we need to make sure that they are responsible, that they are absolutely clear, that if things go wrong, if the agents go rogue, to use the metaphor, then they know that they are on the hook.”
