Britain should expect a ransomware attack involving AI-powered malware on its critical national infrastructure by 2028. That is the central judgment of the United Kingdom National Cyber Security Centre (NCSC) in its annual State of the Cyber Nation report, published on Tuesday 3 November 2026, and it is the starkest official warning yet about what artificial intelligence means for the systems that keep a country running.
In short: the NCSC judges an AI-powered ransomware attack on UK critical infrastructure highly likely by 2028, because AI lets attackers exploit long-tolerated flaws at unprecedented scale. This article explains the warning, the numbers, the actors named, and what to fix first.
The warning is a forecast, not a report of an attack that has already happened. The NCSC judges it highly likely that AI-powered malware will feature in a ransomware attack on critical national infrastructure between now and 2028. In the agency’s language, highly likely is a strong probability judgment, not a certainty and not a sighting. What makes it matter is the reasoning behind it: AI will not necessarily invent new ways in, but will let attackers find and exploit weaknesses that organisations have tolerated for years, at a speed and scale that was previously impossible.
Key takeaways
- The NCSC judges it highly likely that AI-powered malware will be used in a ransomware attack on UK critical national infrastructure by 2028.
- The danger, in the agency’s words, is not new attack methods but old vulnerabilities exploited at unprecedented scale.
- China is named as the most sophisticated and sustained cyber challenge facing the UK; Russia and Iran are also singled out.
- The UK government is funding a national Cyber Shield, a new vulnerability centre and an AI cyber security centre in Manchester.
- The NCSC’s prescription starts with basics: disconnect unneeded systems from the internet, patch promptly, and treat Cyber Essentials as a floor, not a ceiling.
What the NCSC actually warned
The NCSC is part of GCHQ, the UK signals intelligence agency, and its State of the Cyber Nation report is its annual assessment of the country’s cyber health. This year’s report, covering the period from 1 September 2025 to 31 August 2026, says that within the next two years, adoption of AI by hostile actors is highly likely to increase the scale and efficiency of their operations, including ransomware attacks, social engineering and vulnerability exploitation.
Launching the report, NCSC chief executive Dr Richard Horne said the AI revolution must force organisations to reckon with the state of their cyber security, because new technologies will expose vulnerabilities, failings and underinvestment at unprecedented speed and scale. He also framed AI as an opportunity for defenders, while warning that attackers get the same tools and move first against the weakest targets.
“In the coming months, we face a real prospect of an unprecedented rise in the number of cyber incidents. But we must be clear that is not because AI has invented new ways to attack us. But because vulnerabilities that have been tolerated for far too long can now be discovered and exploited at a scale that was once unimaginable.”
The report describes the UK as being at a point of enormous consequence for cyber security, and says the country’s cyber resilience is not yet where it should be. Horne called the moment a choice: invest with urgency now, or explain later to investors, employees, customers and suppliers why food is not on shelves, lights are not on, or businesses have gone under.
Why AI changes the maths for attackers
The NCSC’s argument is economic as much as technical. Finding a vulnerability in a large, old codebase used to be slow, skilled work. AI models can now scan code, probe networks and draft malicious code far faster, which lowers the cost of attacking and raises the return. That matters most where the target estate is old and poorly maintained, because the flaws are already there, waiting to be found.
Horne’s bluntest line was about how those estates got that way. “For decades, we have collectively got away with doing tech on the cheap,” he said. In an interview with the BBC ahead of the launch speech, he condemned decades of underinvestment in operational technology, the systems that control physical processes such as electricity networks, water supplies, switches and refrigeration units. Many organisations, he said, have run equipment beyond its intended lifespan, failed to upgrade it, and failed to isolate it properly from the internet.
That combination is the mechanism behind the 2028 judgment: AI does not need a brand new class of vulnerability, only to find known and neglected ones faster than defenders patch them, and exploit them across many targets at once. The NCSC expects AI to assist several stages of an attack: scouting targets, finding weaknesses in networks and writing malicious code.
The numbers behind the warning
The report’s incident figures tell a mixed story, and the NCSC itself cautions against reading too much into a single year’s movement. Between September 2025 and August 2026, the agency received 1,831 reports of cyber attacks on UK organisations, up from 1,727 the previous year. It gave hands-on support in 337 cases, down from 429, and classed 80 as nationally significant, down sharply from 204 the year before.
| Measure | 2025 to 2026 | Previous year |
|---|---|---|
| Reports of cyber attacks received | 1,831 | 1,727 |
| Cases receiving hands-on NCSC support | 337 | 429 |
| Incidents classed as nationally significant | 80 | 204 |
The fall in nationally significant incidents should not be read as the threat receding. The NCSC notes that incident numbers rarely move in a straight line, and past peaks were often driven by a small number of severe events. Separately, Horne said in a June 2026 lecture that the agency had managed more than 200 incidents affecting critical national infrastructure and its supporting ecosystem in the year to May 2026, with around three quarters believed to be linked to state actors. That figure covers a different period and a narrower set of targets than the annual totals above, so the two should not be added together.
The economic stakes are large. Government research cited in the review puts the cost of malicious cyber activity to UK organisations at about 14.7 billion pounds a year, roughly the size of the UK cyber security sector itself. The review uses the 2025 attack on Jaguar Land Rover as its main example of economic harm: the carmaker reported about 196 million pounds in direct costs, and the review links the incident to UK growth of just 0.1 percent in the third quarter of 2025, below forecasts. The NCSC does not attribute that attack to AI.
Who the NCSC names
The report is unusually direct about state actors. It says China is highly likely to have the intent and be best positioned to use AI for operations against UK critical national infrastructure and other strategic targets, and describes Chinese activity as the most sophisticated and sustained cyber challenge the UK faces. Reuters, reporting the launch, quoted Horne as saying China was likely the best equipped to conduct such operations.
Russia and China are both expected to use AI more routinely in cyber operations by 2028, according to the report. Horne was due to say that Russia was making sustained attempts to penetrate UK military networks, while Iran was carrying out malicious cyber operations globally, including against the UK and individuals seen as a threat.
The supermarket attacks that framed the announcement
The warning landed a day after a wave of cyberattacks on Monday 2 November targeted supermarket freezers across Europe, with temperatures remotely raised to as high as 50 degrees Celsius, forcing stores to discard large amounts of spoiled food. The attacks affected supermarkets in Belgium and Luxembourg as well as Sweden, Ireland, the Netherlands, Estonia and Finland, and followed attacks targeting supermarkets and food suppliers in France in early October. Horne said there had also been some instances in the UK.
Crucially, the NCSC has not attributed the freezer attacks to any actor. Asked whether Russia or Iran might be behind them, Horne said the motive remained unclear. “At this stage, we just don’t know. It might be states. It might be criminals,” he said. The episode is nevertheless exactly the kind of incident the report warns about: an attack on operational technology with a direct physical consequence, empty shelves, rather than stolen data. Whatever the motive, the freezers warmed up and the food was lost.
What the UK is spending in response
Alongside the warning, the government announced new investment in defensive capability. The Chancellor announced 60 million pounds for the first phase of a national Cyber Shield, to develop agentic cyber defence capabilities in partnership with the UK cyber security industry, and 70 million pounds for a new National Exposure and Real-Time Vulnerability (NERV) Centre to identify vulnerabilities across critical infrastructure and the wider economy. A new AI Cyber Security Centre of Excellence, run by the NCSC in Manchester, will develop AI security expertise within government.
The NCSC also pointed to the Cyber Security and Resilience Bill and the forthcoming National Cyber Action Plan as part of the same push. The funding is modest next to the 14.7 billion pound annual cost of malicious cyber activity, and the NCSC is clear that government spending alone cannot close the gap. Its report calls on organisations to make significant changes themselves.
What organisations should do now
Horne’s prescription is deliberately unglamorous, and that is the point. The NCSC wants organisations to fix the basics that AI will otherwise punish first. The measures he named include investing adequately in technology, not connecting technology to the internet unnecessarily, and making Cyber Essentials, the UK government’s baseline certification scheme, the starting point for cyber security rather than the end goal.
For businesses outside critical infrastructure, the same logic applies at smaller scale. The practical checklist drawn from the NCSC’s advice is straightforward: keep systems patched and within supported life, remove internet exposure from systems that do not need it, keep offline backups that ransomware cannot reach, enable automatic updates, and use passkeys wherever possible. Horne’s advice to households was the same in miniature: keep devices updated, maintain offline backups and use passkeys.
The lesson of the report for any organisation, in the UK or elsewhere, is that AI has shortened the time between a vulnerability becoming known and being exploited at scale. Security work that was previously deferrable, replacing an end-of-life system, segmenting a network, patching a known flaw, now competes with an attacker that can find the gap in hours. The two-year window to 2028 is best read as a deadline for that backlog, not as a prediction that nothing will happen before it. For wider context, see our earlier coverage of the tampered Claude Code extension that planted a backdoor in Cursor, the shared robots.txt signal for opting out of AI training, and the Binance hack and what users should do after a major breach.
Frequently asked questions
Has an AI-powered ransomware attack on UK critical infrastructure already happened?
Not according to the NCSC. The report is a forecast: it judges such an attack highly likely by 2028. It does not say one has already occurred, and no specific malware family, company or attack is named in the announcement.
What counts as critical national infrastructure?
The UK definition spans the services a country cannot function without, including energy, transport, defence, food, health and finance, as well as water and communications. The NCSC’s warning covers that whole estate, not just government computer networks.
Why 2028?
2028 is the outer edge of the NCSC’s forecast window, building on an earlier assessment from June 2026 that AI-enabled capabilities would likely be used at scale by then against known vulnerabilities in legacy technology. It is a planning horizon, not a countdown to a known attack.
Does the warning apply outside the UK?
The assessment is about the UK, but the mechanism is universal: any organisation running old, internet-exposed operational technology faces the same AI-assisted discovery problem.
What should a small business do first?
Start with the NCSC’s basics: patch promptly, keep devices updated, remove internet exposure from systems that do not need it, keep an offline backup, and adopt Cyber Essentials as a floor. Those measures address the tolerated vulnerabilities the report says AI will find first.
The bottom line
The NCSC’s warning is less dramatic than some headlines suggest, and more serious. Nobody has announced an AI ransomware attack on a power station or a hospital. What the UK’s cyber agency has said, in unusually plain language, is that the conditions for one are already in place: old systems, thin investment, unnecessary internet exposure, and an attacker toolkit that finds those gaps at machine speed. The era of doing tech on the cheap, in Horne’s phrase, is over because the discount has stopped working. Organisations have roughly two years to act like it, and the cheapest time to fix a tolerated vulnerability is before an AI-assisted attacker finds it first.
Sources and Further Reading
- National Cyber Security Centre, Cyber chief: UK facing AI-powered attacks on critical infrastructure within two years, 3 November 2026 (primary source: report announcement and Horne speech extracts).
- Reuters, UK must expect AI-powered attack on critical infrastructure within two years, report says, 3 November 2026.
- The National, UK cyber chief warns upgrade your tech defences after supermarket hacks, 3 November 2026.
- Insurance Business, NCSC warns of unprecedented rise in cyber incidents, even as serious breaches fall, 3 November 2026.
- TechMarketView, NCSC sees fewer serious incidents as AI exposes weak foundations, 3 November 2026.
- AI Weekly, NCSC warns AI ransomware likely to hit UK infrastructure by 2028, 3 November 2026.
