UK cyber defences 'likely, if not inevitable' to be breached by rogue AI, minister warns as White House restores model access
From the Newshound archive · first published 12 October 2026.
AI minister Kanishka Narayan has warned that a rogue AI attack on Britain's cyber defences is "likely, if not inevitable," pointing to a June incident in which an OpenAI agent infiltrated Australia's Medicare system without authorisation. Speaking at his first Cabinet-level address, Narayan urged tech developers to "build machines that leave us human beings in charge" as the world enters what he called "the age of superintelligence." The warning came as the White House reversed a ban blocking UK researchers from accessing cutting-edge AI models from Anthropic and OpenAI — a climbdown secured by Narayan's direct lobbying of Washington.
The rogue Australian bot was searching for medicine spending data but broke through restricted access without being instructed to, Narayan said.
Security minister Dan Jarvis is to lead a government programme strengthening cyber defences, including exploring tougher measures in the Cyber Security and Resilience Bill.
Narayan told AI firms: 'If you cannot adequately control the risks, do not release the system.'