A dated judgement, in the agency’s own language
The UK’s National Cyber Security Centre published its Annual Review 2026 on 3 November, and the document carries a dated judgement about AI in its formal probability language. “It is also highly likely that between now and 2028, AI-powered malware will have been used in a ransomware attack on UK CNI,” it says, using the standard abbreviation for critical national infrastructure. A second key judgement in the AI chapter repeats it more plainly: “It is highly likely AI malware will be used in a ransomware attack on UK CNI.”
The review pairs that with a state-actor forecast at the same confidence level. “By 2028, it is highly likely China, Russia, and other nation states will routinely use AI cyber to achieve their geopolitical goals with the UK facing an increase in scale and speed of sophisticated cyber attacks,” it says. Separately, it judges that China’s capability programmes “will almost certainly allow China to improve the identification of vulnerabilities, increase targeting efficiency and enhance access operations against critical national infrastructure and other strategic targets.”
Faster, not newer
The agency is explicit that it does not expect AI to produce unfamiliar attacks. “The greatest impact of AI will be the acceleration and scaling of existing tactics, techniques and procedures,” the review says, “as opposed to generating new attack vectors.” It judges it almost certain that AI will act as a force multiplier, and highly likely that AI will keep being folded into threat actors’ operations, “increasing the speed, scale and efficiency of existing attack methodologies rather than autonomously conducting end-to-end attacks.” Human operators, it adds, highly likely remain essential for directing and validating what the models produce.
That framing shapes how the NCSC describes the problem. In the accompanying statement, chief executive Richard Horne said: “For decades, we have collectively got away with doing tech on the cheap.” The review’s own argument is that the weaknesses attackers will exploit are already understood, and that new technology changes the speed and scale at which they can be reached.
One of the review’s AI judgements points inward rather than outward. It is highly likely, the agency says, that the use of “shadow AI” — staff moving company data onto personal devices to use unapproved tools — will leave an organisation’s security open to exploitation by advanced persistent threat groups and criminals, and almost certain that a widening divide will open up where security tasks are frequently handed to shadow AI.
The year’s casework
The incident management team received 1,831 reports of cyber attacks over the year. After triage, 337 incidents required the team’s support, and 80 of those were classed as nationally significant. Through the NCSC’s Early Warning service, the team notified 8,500 organisations of activity suggesting their systems had been compromised.
The review also prices one incident. The attack on Jaguar Land Rover produced approximately £196 million in direct cyber-related costs, the document says, and had a direct effect on GDP, contributing to the UK economy expanding by only 0.1% in the third quarter of 2025.
What to watch
The NCSC’s statement names two funded programmes against this backdrop: £60 million for the first phase of a national Cyber Shield, and £70 million for a National Exposure and Real-Time Vulnerability centre. Whether either is in place before the 2028 window the review sets out is the measurable question.
