Britain’s National Cyber Security Centre (NCSC) says it is highly likely that AI-powered malware will be used in a ransomware attack on the UK’s critical national infrastructure by 2028. The judgment appears in an assessment published on November 3 alongside the NCSC’s annual review, which covers the 12 months to August 31.
The assessment also judges it highly likely that by 2028 Russia, China and other states will routinely use AI in cyber operations to pursue geopolitical goals. It calls it almost certain that many systems will remain vulnerable to AI tools that find and exploit weaknesses, and says attackers will take less care to avoid losing access to a network because replacing that access will cost them little. It judges it likely that attackers will turn AI on known vulnerabilities in older infrastructure technology and on new openings created as operators deploy AI tools.
The review’s chapter on AI says the technology’s main effect will be to speed up and scale existing attack techniques rather than create new ones, and that human operators will highly likely remain essential for directing and checking AI output. It judges that losing control of an agentic AI system during an operation would highly likely hinder attackers from reaching their objectives. China, it says, highly likely has the intent and is best positioned to use AI against critical infrastructure and other strategic targets.
The assessment sets out three scenarios for 2028, which the NCSC describes as plausible but not predictions. In the optimistic one, AI development slows and critical infrastructure operators replace aging systems. In the pessimistic one, a new threat has emerged: an automated attacker using frontier AI that is no longer under its operator’s control. The NCSC calls it “a realistic possibility” that such an entity could “self-replicate and conduct cyber attacks at will, with no apparent motivation or restraint.” In October, three US agencies published practices for responding to incidents in which AI agents take significant actions they were not instructed to take.
The annual review shows that most successful attacks still use familiar methods, such as known vulnerabilities in internet-facing systems, stolen credentials and social engineering. The NCSC received 1,831 reports of cyberattacks during the year, of which 337 needed its support and 80 were classed as nationally significant. Its incident team has begun to see attackers use AI tools, but none of the uses it observed enabled a new class of attack.
“AI did not create these cyber security weaknesses; it revealed them,” Richard Horne, the NCSC’s chief executive, wrote in the review’s foreword. In remarks prepared for the report’s launch, he warned of “a real prospect of an unprecedented rise in the number of cyber incidents” in the coming months, because long-tolerated vulnerabilities can now be found and exploited at far greater scale.
The government announced last week £60 million for the first phase of Cyber Shield, a program in which the NCSC will supply security data to AI defensive agents operating inside UK organizations, and £70 million for a new center to identify vulnerabilities across critical infrastructure and the wider economy.
Sources: NCSC assessment, NCSC Annual Review 2026, NCSC announcement
–
By the Control Plane Editorial Team