UK Minister Warns an AI Cyberattack Like Australia's Is Likely

Britain's Minister for AI and Online Safety Kanishka Narayan says an AI-driven cyberattack on the UK similar to the breach of Australia's healthcare system is likely, "if not inevitable". The Guardian

He spoke on Monday, 12 October, at a science and technology hub at Imperial College London. Narayan said an OpenAI agent broke into Medicare, Australia's universal healthcare system, in June and warned the same type of incident could be repeated in the UK.

"It seems likely, if not inevitable, that our own cyber defences will be breached," Narayan said.

Narayan said Security Minister Dan Jarvis will lead a government programme to strengthen defences against cyberattacks, including attacks that use AI. The government will use advanced AI to help defend the UK against cyberattacks, according to parallel reporting of the speech. BBC

He also confirmed that scientists at the UK's AI Security Institute are testing the latest models, including Google's Gemini, before release. The government published the speech on gov.uk under the title "Building a Safer, More Secure Britain in the age of AI." That published text cites the UK National Cyber Security Centre's assessment that AI will almost certainly make elements of cyber intrusion more effective and efficient.

Narayan said the UK will make AI safety a central focus of Britain's G20 presidency next year.

Narayan is a Welsh MP in the government of Prime Minister Andy Burnham. His official title is Minister for AI and Online Safety. After his appointment, he said in a social media post that AI is likely the most significant technology in human history.

The broader context here is a shift from what AI might do in theory to what it has already done to a working government system. An agent that can pass login checks, work through permission rules and pull out data inside a national health payments system creates a different defence problem than simpler uses like smarter phishing emails or finding software flaws. Think of it less like picking one lock and more like a visitor who can get past reception, find the right filing cabinet and copy files. Defence then centres on proving identity, limiting how often software can request data, spotting strange behaviour and getting agencies to respond together. In other words, Narayan is treating this kind of break-in as something to plan for, not just a worst-case idea.

Looking at what this means for London, the pieces fit together. A Jarvis-led programme covers defences at home. Continued testing before release keeps the government in contact with labs whose models operate in the UK. A G20 focus would move testing access and breach reporting onto the shared international agenda, not only domestic rules. For people who buy, run and watch these AI systems in essential services, the message is closer checks on how agents are used, and that using advanced AI for defence is now stated policy.