The National Cyber Security Centre warns an AI-powered ransomware attack on Britain's critical infrastructure is coming within two years. Its own report says the scenario behind that warning is explicitly not a prediction.
Estimated reading time: 6 minutes
In short. The National Cyber Security Centre's new assessment states, on its first page, that the three scenarios it sets out "are assessed to be plausible possibilities in the medium-term but are not predictions" [1]. On its second page, under a heading with no such qualifier, the same document states as a stand-alone key judgement that "it is highly likely that between now and 2028, AI-powered malware will have been used in a ransomware attack on UK CNI" [1]. The press release launching the report, published the same day, drops the scenario framing and the "not predictions" line entirely [2]. That release also points to last week's Budget as evidence the government is already responding, but the Budget's own headline figure for that response has already been shown, by this site's own direct check published yesterday, to be £176 million higher than its own itemised total [3]. Two documented mistakes: four stars.
GCHQ's National Cyber Security Centre published a new threat assessment on 3 November 2026, Artificial intelligence and cyber intrusion, alongside a speech from its chief executive, Dr Richard Horne, warning that Britain's "tech on the cheap" habit has left critical infrastructure exposed to AI-accelerated attack [2]. The report itself is careful about what kind of document it is. The press release built to launch it, and read by many more people than the report itself, is not.
The claims, tested
| The document's own words | What the record shows | Verdict |
|---|---|---|
| NCSC report, page 1, "Context": "This assessment examines current AI-enabled threat trends and projections of the threat landscape to 2028 through three scenarios. The scenarios are assessed to be plausible possibilities in the medium-term but are not predictions" [1] | The same report, page 2, under "Key judgements", with no scenario qualifier attached: "It is highly likely that between now and 2028, AI-powered malware will have been used in a ransomware attack on UK CNI." The launch press release drops the hedge further still: "AI-powered malware highly likely to be used in ransomware attack on Britain's critical national infrastructure by 2028, new report warns" [1] [2] | The report's own disclaimer that it is not predicting anything does not reach the sentence that becomes the headline |
| NCSC press release: "last week the Chancellor also announced greater investment in capabilities to help the UK harness AI for cyber defence, including £60 million into developing the first phase of a national Cyber Shield, £70 million for a new National Exposure and Real-Time Vulnerability (NERV) Centre... and a new AI Cyber Security Centre of Excellence" [2] | Budget 2026's own text states the government "will therefore invest around £500 million to build the secure foundations for an AI-enabled economy", but the paragraphs that figure introduces itemise only £314 million; the remaining £176 million appears only in a Treasury scorecard line whose own title says it also funds AI adoption and capability-building, spending the Budget's later pages treat as a separate stage [3] | The investment NCSC cites as the government's response is real, but the headline total behind it overstates the itemised security spend by more than a third |
A disclaimer that stays on page one
The report's structure is, on its own terms, a model of hedged forecasting. It sets out three named scenarios, baseline, pessimistic and optimistic, built from five identified drivers (geopolitical tensions, UK critical national infrastructure cyber defence, the pace of AI development, the availability of AI-cyber capability, and workforce cyber literacy), and states plainly that "the prospect for surprise is high" [1]. Every substantive sentence in the body carries a formal probability term, "highly likely", "almost certain", "likely", or "realistic possibility", drawn from the Professional Heads of Intelligence Assessment's standard yardstick, the same discipline used across government threat assessment [1]. That is a genuinely more careful document than most threat forecasting put out under a government crest.
The "Key judgements" on page 2, though, sit before the three scenarios are even introduced, and carry no scenario label of their own [1]. The most newsworthy of the four, that AI-powered malware will "highly likely" have been used in a UK ransomware attack by 2028, is not flagged as contingent on the pessimistic scenario rather than the optimistic one, where "UK CNI has an incredibly robust defence against any, and all, cyber attacks from a range of actors using AI-cyber tools" [1]. A reader who stops at the key judgements, which is most readers, gets a single confident forecast. A reader who reaches page 1's "Context" section first would know that forecast is explicitly not what the document claims to be offering.
The press release goes one step further than the report itself. Its headline is "Cyber chief: UK facing AI-powered attacks on critical infrastructure within two years"; its second paragraph states "AI-powered malware highly likely to be used in ransomware attack on Britain's critical national infrastructure by 2028, new report warns" [2]. Neither sentence carries the word "scenario". Neither mentions that the report explicitly disclaims prediction. Dr Horne's quoted speech reinforces the flattened version: "In the coming months, we face a real prospect of an unprecedented rise in the number of cyber incidents," he is reported as saying, a claim about the near future stated without qualification, in a release that otherwise reproduces the report's own figures accurately [2].
The funding the release points to
The press release does not rest only on the warning. It points to money: "last week the Chancellor also announced greater investment in capabilities to help the UK harness AI for cyber defence, including £60 million into developing the first phase of a national Cyber Shield, £70 million for a new National Exposure and Real-Time Vulnerability (NERV) Centre to identify vulnerabilities across critical infrastructure and the economy, and a new AI Cyber Security Centre of Excellence, run by the National Cyber Security Centre, in Manchester" [2]. Those three figures are accurate: Budget 2026, published 28 October 2026, does commit £60 million to Cyber Shield and £70 million to the NERV centre, each with a named host and a stated purpose [3].
But the press release's framing borrows the Budget's own umbrella claim, that the government is investing at the scale the warning implies is needed. Budget 2026's paragraph 4.129 states the government "will therefore invest around £500 million to build the secure foundations for an AI-enabled economy" [3]. This site's teardown of that section, published yesterday, found that the paragraphs 4.129 itself introduces, 4.130 and 4.131, itemise five costed measures totalling £314 million: £120 million for the AI Security Institute, £70 million for the NCSC exposure centre, £24 million for public sector AI defence, £60 million for Cyber Shield, and £40 million for a new National Agentic Security Unit [3]. The £176 million gap between that total and "around £500 million" closes only in the Treasury's Table 6.1 policy-decisions scorecard, whose own title for the matching line states it funds AI security, AI adoption and AI capability-building together, not security alone, spending the Budget's own paragraph 4.134 introduces separately, under a different heading, as a later and distinct stage [4]. None of that makes the £60 million or £70 million NCSC cites untrue. It means the implicit reassurance, that the government's response matches the scale of a "highly likely" national infrastructure attack, rests on a headline figure this site had already tested and found inflated by more than a third, a day before NCSC's own release pointed back to it.
Credit where due
The report itself is unusually disciplined for a government cyber assessment. It names its own methodology (the PHIA probability yardstick), lays out an optimistic scenario alongside the pessimistic one rather than forecasting doom alone, and is explicit, in writing, about the limits of what scenario planning can claim to know. The specific funding figures in NCSC's press release match the Budget documents exactly, not an inflated or rounded version of them. And the report's "Possible drivers" section names UK critical infrastructure's own cyber defence investment as a variable the UK can still act on, not a fixed condition, which is a more honest framing than a straight doom warning would offer.
The star score counts two documented mistakes. First, the report's own page 1 disclaimer that its three scenarios "are not predictions" does not carry through to the "Key judgements" on page 2, where the single most newsworthy claim, a "highly likely" AI-powered ransomware attack on UK critical national infrastructure by 2028, is stated without a scenario qualifier; the launch press release and the NCSC chief executive's quoted speech drop the hedge further, presenting a flat warning with no reference to scenarios or to the report's own disclaimer. Second, the press release cites last week's Budget investment as evidence of government action at the scale its warning implies is needed, but the Budget's own headline total for that investment, "around £500 million", is £176 million above the £314 million its own itemised breakdown actually costs, a gap this site's own Budget 2026 teardown verified directly from the Treasury's own documents the day before this release pointed back to that figure. This piece finds no wrongdoing by the National Cyber Security Centre or any named individual; Dr Richard Horne is quoted only in his capacity as NCSC chief executive, from the agency's own published release. Both NCSC pages and the Budget documents were fetched directly on 3 November 2026 and returned HTTP 200.
Sources
- Artificial intelligence and cyber intrusion (PDF), National Cyber Security Centre, published 3 November 2026
- Cyber chief: UK facing AI-powered attacks on critical infrastructure within two years, National Cyber Security Centre, published 3 November 2026
- Budget 2026 (HTML), HM Treasury, 28 October 2026
- Table 6.1 and 6.2 Budget 2026 Policy Decisions (spreadsheet), HM Treasury, 28 October 2026