Britain’s National Cyber Security Centre expects AI-powered malware to feature in a ransomware attack on the country’s critical infrastructure by 2028, Quartz reported on November 3. The warning, issued in the agency’s annual review, is a forecast rather than an announcement that such an attack has occurred.

The central concern is speed and scale, not entirely new forms of hacking. NCSC chief executive Richard Horne said attackers could use AI to identify and exploit weaknesses that organizations have left unresolved for years. The review expects AI primarily to amplify established attack techniques.

Coverage published by Euronews described the physical systems at stake: electricity controls, water facilities, dams, manufacturing equipment and refrigeration. Horne pointed to recent attacks on French supermarket refrigeration systems as an example of how vulnerable technology can disrupt everyday services.

Those refrigeration attacks did not involve AI, Horne said. The systems were exposed to the internet, used default passwords and depended on outdated technology that no longer received updates. He told The Associated Press that refrigeration systems in Britain had also been affected and that NCSC specialists had contacted organizations using vulnerable equipment.

The review identifies China as likely having both the motivation and the strongest capability to deploy AI against critical infrastructure. It also anticipates routine AI use by China, Russia and other states to advance geopolitical objectives by 2028. Quartz noted that Beijing, Moscow and Tehran have repeatedly denied comparable allegations of hostile cyberactivity.

The agency urged organizations to address basic security gaps now. Its recommendations include adopting the government-backed Cyber Essentials standard, keeping systems updated and restricting administrator access. The review cites government research estimating annual losses from malicious cyberactivity to U.K. organizations at £14.7 billion.