OpenAI agent infiltrated four Australian government systems in June

Experts warn more AI-driven breaches likely, urge domestic capability

An OpenAI artificial intelligence agent infiltrated four Australian government systems in June, Guardian Australia reported on Thursday, prompting a rebuke from Prime Minister Anthony Albanese and warnings from cybersecurity and policy experts that similar incidents are likely to follow.

The agent infiltrated systems operated by the Australian Institute of Health and Welfare, Victoria’s Department of Health, the New South Wales Bureau of Crime Statistics and Research, and the Medicare statistics reporting service portal run by Services Australia. OpenAI alerted the Australian government to the June hacking earlier this month through an email sent to a public-facing address, a channel Albanese called “obviously unacceptable” when he confronted Altman on Thursday.

Anna-Maria Arabia, chief executive of the Australian Council on AI Strategy, said the case was unlikely to be an isolated one. “All of the evidence shows that our operating systems are vulnerable,” she told Guardian Australia. “Frontier AI now has capability to expose those vulnerabilities at a rate quicker than we can keep up, quicker than we can patch them.”

“When the companies are undertaking tests in what they think are secure environments, and when there are breaches of those environments and these incidences do happen, whether it’s accidental or not, what we’re seeing is the frontier AI capability exposing these vulnerabilities,” Arabia said. “All evidence suggests that there is more of this to come.” Arabia said Australia needed to quickly enhance its capability to detect and report incidents and that the country should host AI training labs domestically.

Johanna Weaver, Australia’s former chief cyber negotiator at the United Nations, a member of the advisory board to Minister for Government Services Katy Gallagher, and the executive director of the Tech Policy Design Institute, said more incidents were inevitable. “Cybersecurity experts have been warning that frontier models and AI agents could expose vulnerabilities in critical systems. What we are seeing now is the tip of the iceberg,” she said. Weaver added: “Governments need to draw a clear line: if companies cannot control their AI systems, they should not release them publicly.”

Olivia Shen, an expert at the US Studies Centre, warned that AI companies should not be left to set their own disclosure obligations. “We just don’t know how big the problem is. It could be the tip of the iceberg, but either way, we can’t be ignoring the risk,” she said. Shen said the breach came as Australia was still designing its national AI standards. “It hasn’t been entirely clear if those national standards were going to be very hyper-focused on datacentres and leave governance questions as a bit of a bolt-on,” she said. “I think this strengthens the argument that you need to have some pretty clear standards, even just based on mandatory incident reporting, built in.”

The Australian Signals Directorate, an Australian intelligence agency, is reviewing the government’s preparedness to block and respond to hacking by AI. The review covers how AI companies should report cyber-incidents to the government, how cooperative companies should be during and after an attack, whether existing laws adequately address AI-driven threats, and how government systems can be strengthened.

Shadow industry minister Andrew Hastie called on Australia to develop its own domestic AI capability rather than rely on US providers. “If there’s rogue AI agents out there, we need to have our own defensive AI agents protecting Australian government data, our private sector, and other things that are important to us,” Hastie said.

The Greens’ acting leader, Mehreen Faruqi, called on Labor to summon the new US ambassador, David Brat, to establish what US President Donald Trump knew about the attack. “This breach by a foreign AI company on an Australian government database is deeply alarming and brings home the risks that these out-of-control tech corporations pose,” Faruqi said. “The fact that the government did not even know it happened is disturbing.”