The UK’s National Cyber Security Centre (NCSC) judges it highly likely that AI-powered malware will be used in a ransomware attack on UK critical national infrastructure between now and 2028. The judgement is in its assessment “Artificial intelligence and cyber intrusion”, published on 3 November 2026. Operators should treat it as a planning input, not a date. The checklist at the end is our advice, not NCSC guidance.
What Does the NCSC AI Malware Assessment 2028 Say?
Each NCSC likelihood word sits next to its claim. The last row comes from the NCSC news item; the rest come from the assessment page.
| Source | Judgement | NCSC wording |
|---|---|---|
| AI and cyber intrusion | AI-powered malware will have been used in a ransomware attack on UK CNI between now and 2028 | Highly likely |
| AI and cyber intrusion | By 2028, Russia, China and other nation states will routinely use AI-cyber | Highly likely |
| AI and cyber intrusion | By 2028, many systems will remain vulnerable to attack from AI-cyber tools | Almost certain |
| AI and cyber intrusion | By 2028, attackers will apply AI-cyber capability against known vulnerabilities in critical national infrastructure | Likely |
| NCSC news item (State of the Cyber Nation) | AI adoption by hostile actors will increase the scale and efficiency of their operations within the next two years | Highly likely |
The assessment sets out three scenarios for 2028 and calls them plausible possibilities, not predictions. In the baseline the global threat increases. In the pessimistic case capability spreads faster than UK defences, and in the optimistic case development slows.
Which Actors and Weak Points Does the NCSC Name?
In its pessimistic scenario, the NCSC describes China as a leading developer of AI-cyber capability that has likely played a role in passing advanced capability to other actors. In its baseline 2028 scenario, the NCSC says it is highly likely that Russian cyber actors have adapted and retrained open-source AI models. It says this has highly likely led to more pre-positioning, espionage and intellectual property theft against UK interests.
On weak points, the assessment says that by 2028 attackers are likely to apply AI-cyber capability against known vulnerabilities in legacy technology in critical infrastructure. Its key judgement points to new opportunities in the attack surface from deploying AI tools inside CNI. In its baseline scenario, defensive capabilities in UK CNI lag behind offensive ones.
The NCSC news item says Dr Richard Horne, its chief executive, is expected to speak at the launch of the State of the Cyber Nation report. He is expected to say the risk is not because AI has invented new ways to attack. It is because “vulnerabilities that have been tolerated for far too long can now be discovered and exploited at a scale that was once unimaginable.”
What Does the NCSC Tell Organisations to Do?
The assessment page has no recommendations section of its own. The NCSC news item says Dr Horne will urge organisations to address under-investment in technology and to avoid connecting technology to the internet unnecessarily. It also says to make Cyber Essentials the starting point, not the end goal.
A Readiness Checklist for UK Critical Infrastructure Operators
- List what is old and exposedBuild an inventory of legacy systems and of every service reachable from the internet, then remove any connection that is not needed. This shrinks your attack surface.
- Fix known flaws firstRank open vulnerabilities by exposure and by whether they are being exploited, and patch internet-facing ones first. If a system cannot be patched, isolate it.
- Separate what must not failSegment operational technology from office networks, and keep the paths between them few and monitored.
- Plan for AI-driven ransomwareKeep offline backups, test a restore, and rehearse the ransomware response playbook with operations staff.
- Watch AI inside your own estateRecord where AI tools and agents are deployed, since the NCSC points to new attack surface from AI tools in CNI. Our analysis of an AI-agent ransomware case shows the defender side.
Our glossary explains critical infrastructure. For how attackers already use AI frameworks, read our explainer on CyberStrikeAI attacks.