UK's NCSC: AI malware "highly likely" in a ransomware attack on critical infrastructure by 2028

The UK's National Cyber Security Centre (NCSC) published its 2026 Annual Review on November 3, 2026, judging it "highly likely" that AI-powered malware will be used in a ransomware attack on UK critical national infrastructure by 2028. Its core message is that AI mostly speeds up attacks that already work — exploiting old, unpatched weaknesses — so organisations that patch slowly and run legacy systems are most at risk. The practical step: plan now for a much faster patching cycle.

What happened, in plain English

The NCSC is part of GCHQ, the UK's signals intelligence agency, and acts as the country's technical authority for cyber security. Every year it publishes a review of the threats it has seen. This year's edition, titled The State of the Cyber Nation, covers September 1, 2025 to August 31, 2026, and places AI at the centre of its forecast.

The headline judgment concerns critical national infrastructure (CNI) — energy, water, transport, health and other services the country depends on. The NCSC says AI-powered malware is "highly likely" to be used in a ransomware attack (malware that locks or steals data and demands payment) against UK CNI by 2028. In the NCSC's intelligence language, "highly likely" means a strong probability, not a certainty, and this is a forecast, not a report of an attack that has already happened.

The review's more important point is less dramatic. The NCSC does not expect AI to invent new kinds of attack. It expects AI to make existing ones faster, cheaper and more numerous — especially finding and exploiting software flaws that "have been tolerated for far too long."

An analogy: AI does not give burglars a new way through walls. It gives them a tool that checks every door and window on the street in minutes. Houses with old locks get found first.

NCSC chief executive Dr Richard Horne put it bluntly: "For decades, we have collectively got away with doing tech on the cheap."

Are you affected? What to do now

If you work in or supply UK critical infrastructure, government or a regulated sector, this review is aimed at you. Everyone else should read it as a well-sourced signal of where attackers are heading. There is no emergency patch or deadline today; the actions are about readiness.

• Prepare for a "patch wave". The NCSC tells organisations to "prepare now for when a 'patch wave' arrives": as AI finds more bugs, vendors will ship more fixes, faster. Check how long your critical patches take today and who approves them. The NCSC notes that UK government cut its median time to fix domain-related vulnerabilities from 50 days to 8 days — a useful benchmark.
• Know your estate. The review warns that organisations with "ageing technology, limited visibility of their technology estates and slow remediation processes" will struggle. Keep an up-to-date asset inventory, including what faces the internet.
• Reduce internet exposure. Horne's advice is to avoid connecting technology to the internet unnecessarily. The review's "7 things we learned" says "internet-facing assets remain a primary route to compromise."
• Treat the basics as a floor. The NCSC calls Cyber Essentials (the UK government-backed baseline certification) "the starting point for cyber security, not the end goal."
• Audit your own AI tools. For agentic AI tools — AI that can take actions in your systems — the NCSC says "organisations need to understand how they are used" and what they can access. It also flags "shadow AI": staff moving sensitive data to personal devices to use unofficial AI tools. Check your acceptable-use policy and data-loss controls.
• Brief staff on social engineering. The NCSC expects AI to increase the scale of social engineering (tricking people into giving access). "People remain the target" is one of its seven lessons.
• Boards: the NCSC asks boards for "measurable improvements in preparedness, recovery capability and resilience outcomes", not just the existence of policies. It points to its Cyber Security Toolkit for Boards and its guidance on recovering from a highly disruptive cyber attack.
• Regulated UK sectors: watch the Cyber Security and Resilience Bill, which the review says will bring a more enforceable framework and new sectors into scope. The review gives no commencement date. (We covered the UK's plans for AI in that law on October 13, 2026.)

The review publishes no indicators of compromise (IOCs); it is a strategic assessment, not an incident alert.

The expert view

The AI chapter contains seven judgments, and their probability language is worth reading carefully. The NCSC says it is "almost certain AI will act as a force multiplier" across reconnaissance, social engineering, vulnerability discovery and exploit writing, while "human operators highly likely remain essential for directing, validating and adapting AI-generated outputs." That matches what threat-intel teams have published this autumn: AI agents doing the labour of intrusion with a human approving steps, and initial access still coming through known weaknesses and stolen passwords.

The most consequential judgment for defenders is about vulnerability discovery. The NCSC says AI-assisted discovery and exploit development will "highly likely" pose a significant threat, and that defenders face "a race to shorten patch deployment timelines." The pressure works in two directions. Defenders and vendors are using the same capability to find bugs — recent industry reports describe AI uncovering flaws at a scale that strains triage. The result is more advisories, more often, and shorter windows between a fix and an exploit. Organisations whose change process assumes a monthly cycle will fall behind.

The second structural point is a "widening digital divide", which the NCSC calls "almost certain". Mature organisations that adopt AI with governance will improve their defences; those with legacy systems and small teams become more attractive targets. In practice that puts small suppliers to critical services in focus — and supply-chain risk was already one of the review's lessons.

On attribution, the NCSC names one state: it says "China highly likely has the intent and is best positioned to use AI for operations," linking that to China's frontier AI programmes and interest in CNI. That is the NCSC's assessment; it cites no specific campaign on the page.

Separately, a chapter called "The age of autonomous attack" sketches a pessimistic 2028 scenario in which an AI-driven "automated cyber actor" is "no longer under its operator's control" and could, as a "realistic possibility", self-replicate. The NCSC says this chapter is not a prediction but an illustration of what inaction could cost. It should be read that way.

What is new here is the government response. The announcement lists £60 million for the first phase of a national "Cyber Shield" — which the NCSC describes as a programme "to build agentic detection, analysis and response", where defensive AI agents in organisations share signals with a national platform, starting with detection and blocking rather than automatic fixes. It also lists £70 million for a National Exposure and Real-Time Vulnerability (NERV) Centre and a new NCSC-run AI Cyber Security Centre of Excellence in Manchester.

What is still unknown: The 2028 date appears in the NCSC's announcement; the AI chapter itself states the judgment without a date. The review gives no figures for AI-enabled incidents so far. In its incident chapter, the NCSC reports 1,831 reports received, 337 incidents needing its support and 80 nationally significant incidents, but does not say how many involved AI. Cyber Shield's later stages, participation rules and dates are not yet published.

Official sources

• NCSC — Annual Review 2026: The State of the Cyber Nation (November 3, 2026) — full PDF
• NCSC — "Cyber chief: UK facing AI-powered attacks on critical infrastructure in two years" (press release)
• NCSC Annual Review 2026 — The impact of AI on cyber threat
• NCSC Annual Review 2026 — The age of autonomous attack (scenario)
• NCSC Annual Review 2026 — Incident management
• NCSC Annual Review 2026 — 7 things we learned
• NCSC Annual Review 2026 — A moment of choice
• NCSC Annual Review 2026 — Cyber Shield
• Coverage: Tech Insider — AI malware will hit UK CNI by 2028, NCSC warns