The UK’s National Cyber Security Centre says AI malware will highly likely be used in a ransomware attack on the country’s critical infrastructure. Here is what it says, what it does not predict and what the government is paying for.
CTRL + ALT News | Artificial Intelligence | Information checked on 8 November 2026
On 3 November 2026 the National Cyber Security Centre (NCSC), the UK’s cyber security agency, published its Annual Review for the year from 1 September 2025 to 31 August 2026. Its chapter on artificial intelligence reaches a stark conclusion: “it is highly likely AI malware will be used in a ransomware attack on UK CNI”. CNI is critical national infrastructure, the networks and services the country depends on. In its news item the NCSC puts a date on it: by 2028, or within two years.
This is a judgement by the UK’s cyber agency, not a measurement or an event that has happened. Reading the pages closely shows both how serious the warning is and where it stops.
What the NCSC actually says
The NCSC uses verbal probability terms such as “highly likely” and “almost certain”. We found no percentages on the pages we read, so how likely “highly likely” is remains the NCSC’s own scale.
Its main points on AI:
- AI is a “force multiplier”. The chapter says AI is “acting as a force multiplier across the cyber threat landscape”, improving several stages of an attack, including reconnaissance, social engineering and the discovery of vulnerabilities. It speaks of the acceleration and scaling of tactics attackers already use.
- It is not predicting robot hackers working alone. The NCSC expects AI to be integrated into attacks “rather than autonomously conducting end-to-end attacks”, and says human operators will highly likely remain essential for directing and checking the work. It also points to the risk of losing control while running an agentic AI, which it says will highly likely hinder attackers.
- China is singled out. The chapter says China “highly likely has the intent and is best positioned to use AI” in operations against critical national infrastructure and other strategic targets.
- “Shadow AI” is a risk. When staff move sensitive company data to personal devices or unapproved AI tools, the chapter says organisations lose visibility of how that information is handled.
Defenders get help too, but not evenly
The review does not see AI as only a threat. It says organisations that adopt it are likely to get better at detecting threats, prioritising vulnerabilities and automating defence. But it warns of a widening digital divide: the benefits will not be shared equally. It also says defenders and system maintainers will face a race to shorten the time it takes to deploy patches.
That race is already visible in the news. Our report on the PHP2Shell flaws shows what happens when serious vulnerabilities are attacked before every site is patched.
What the government is paying for
The NCSC’s news item says the government is putting £60 million into the first phase of a national Cyber Shield and £70 million into a National Exposure and Real-Time Vulnerability (NERV) Centre. It also announces an AI Cyber Security Centre of Excellence, run by the NCSC, in Manchester.
The Register, which covered the review the same day, reports more:
- £40 million for a National Agentic Security Unit and £24 million for public-sector defences.
- A 75% discount on Cyber Essentials certification for micro, small and medium-sized businesses applying for their first certificate.
- A pilot called Project DROMOS, which analysed government code repositories for four weeks and found “hundreds of potential security issues”. The Register describes the sum for the Manchester centre as undisclosed.
We have attributed these to The Register because we could not check them in the full review, which was too large for our tools to open.
The agency’s chief executive, Richard Horne, argues the country has under-invested. “For decades, we have collectively got away with doing tech on the cheap,” he says in the NCSC’s news item, and he urges leaders to invest “with urgency”.
What it means for UK organisations
This is our reading, not the NCSC’s recommendation list. The review’s own themes suggest three practical questions for any organisation:
- Do you know which AI tools your staff use? The shadow AI warning is about visibility.
- How fast can you patch? The NCSC expects the time attackers have to act to shrink.
- Are the basics in place? The NCSC’s news item calls Cyber Essentials the starting point for cyber security, not the end goal.
Recent incidents show the same pressure points. See how India says AI agents reached organisations’ live systems without authorisation, and what happened in the OpenCode breach and how to protect your AI keys.
Frequently asked questions
Has an AI-powered ransomware attack on UK infrastructure already happened? The pages we read do not say so. The NCSC is forecasting that one will highly likely happen, not reporting one.
Does “by 2028” appear in the Annual Review itself? It appears in the NCSC’s news item about the review, and in The Register. It does not appear on the AI chapter page we read.
Will AI hack systems entirely on its own? The NCSC does not expect that. It says human operators will highly likely remain essential.
What we could not confirm
- The full PDF of the Annual Review, which we could not open.
- The funding for the National Agentic Security Unit, the public-sector defences and the Cyber Essentials discount, and the details of Project DROMOS, which come only from The Register.
- What “highly likely” means in numbers, if the NCSC defines it elsewhere.
- Any real-world case of AI malware used in a ransomware attack on UK infrastructure.
Sources: NCSC news item and Annual Review chapter, 3 November 2026; The Register, 3 November 2026 (links in the text above).
Read the NCSC’s pages: news item and AI chapter. Coverage: The Register.
Leave a Reply