AI 'agent' infiltrated Australian Government website in first-of-its-kind breach

An AI agent has infiltrated an Australian Government website in a first-of-its-kind attack. Anthony Albanese, Australia's Prime Minister publicly revealed that an artificial intelligence developed by US firm OpenAI had breached a Government portal. An AI agent is a type of software which is given permission to complete a task on the internet. Anthropic, OpenAI, Meta and Microsoft have all developed AI models to do jobs like replying to emails, scheduling appointments and to order items online. TRENDING Stories Videos Your Say Mr Albanese, the leader of Australia's Labour Party, revealed to the world the breach while attending the UN General Assembly in New York. Nonsensitive medical data and statistics around spending were accessed by the AI, Mr Albanese said. No personal information related to the publicly-funded universal health insurance scheme is believed to be taken, he added. But officials are aware of three other systems "which may be impacted". These include the Australian Institute of Health and Welfare, the New South Wales Bureau of Crime Statistics and Research, and the Victorian Department of Health. This is believed to be the first time an AI has accessed a Government database. He told reporters in New York: "This situation is obviously unacceptable. And today I spoke with the CEO of Open AI, 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 uh what had occurred and the the nature of the way that that notification occurred as well was unacceptable." Mr Albanese said an OpenAI agent which was researching public health spending accessed his Government's portal website on June 18. The sophisticated piece of software was blocked from accessing the spending data it needed by the website so it found alternative ways of accessing the information. The Australian Prime Minister said: "There were blocks clearly which were coming back, telling the AI agent no. The AI agent found a way around those blocks. "Didn't accept no for an answer if you like." LATEST ON AI SAFETY: AI should be treated as species that could disempower us all, whistleblower warns Donald Trump lashes out at 'conspiracy theorists' raising alarm over rapid AI development in 'sick hoax' Prince Harry warns against 'gambling with children's lives' as he calls for AI restrictions OpenAI became aware of the breach in August, but only notified the Australian Government in an email to Services Australia on September 10. Officials from the agency then notified the Australian Cyber Security Centre on September 15, who in turn briefed Mr Albanese just before he took off for the UN. A task force has now been established to investigate the hack, and the "incident will inform the development of our government's AI standards legislation", Mr Albanese said. The announcement came as the issue of AI safety took centre stage at the UN. OpenAI chief Mr Altman addressed the UN Security Council calling for international standards on the technology to ensure it is "democratic" on Wednesday. Andy Burnham used his time on the world stage to announce a new quango, the " National Centre for Information Defence ", to tackle AI-fuelled "hostile information attacks" and "disinformation" from foreign actors. Donald Trump has pushed back against AI safety , saying that fears around its capabilities were a "hoax", similar to the "Global Warming Scam". OpenAI, in a statement, said it became aware of the breach during an "extensive review" of "misaligned model activity". It said: "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. "In the course of that, our models took actions we did not intend." The San Francisco-based firm said it found no evidence of patient data being accessed, but had notified relevant organisations. It is now "providing technical information to support their investigations and help address potential security vulnerabilities". "Our overall review is ongoing, and we remain committed to transparency about these issues and to sharing what we learn as that work continues," it added. Our Standards: The GB News Editorial Charter