What happened?
The National Cyber Security Centre (NCSC), part of GCHQ, has warned that AI-powered malware is highly likely to be used in a ransomware attack on the UK’s critical national infrastructure by 2028. It's State of the Cyber Nation report anticipates that hostile actors’ adoption of AI will increase the scale and efficiency of ransomware attacks, social engineering and vulnerability exploitation over the next two years.
The NCSC’s central concern is that longstanding security weaknesses, underinvestment in technology and unnecessary internet connectivity will leave organisations increasingly exposed. AI could enable attackers to identify and exploit these weaknesses much faster and on a greater scale. NCSC Chief Executive Dr Richard Horne is therefore urging leaders to make security and resilience fundamental business requirements rather than afterthoughts.
So what?
The NCSC warning has implications beyond organisations operating critical infrastructure. Disruption to electricity, communications, food distribution or other essential services could affect the businesses and communities that depend on them. Even organisations with strong internal security could experience operational difficulties if an essential supplier or infrastructure provider were compromised in a domino effect.
The NCSC highlights how accumulated security weaknesses become more dangerous as attackers’ capabilities improve. Unsupported technology, unresolved vulnerabilities and unnecessarily exposed systems may all provide opportunities for exploitation. The significance of AI is not necessarily that it introduces entirely new attack methods, but that it could accelerate established techniques and increase the volume of attacks organisations must withstand.
There are currently no reported AI-led attacks targeting the UK. AI use is limited to creating lures and phishing emails, and the attacks themselves are still carried out by humans. However, AI is developing rapidly, and it is only a matter of time before AI-led attacks become mainstream.
An example of the type of infrastructure attack that could be in scope is the so-called “fridge attacks”, cyberattacks on the remote-management systems of commercial fridges and freezers, which hackers used to push units into defrost mode and spoil food. A first wave hit around 700 French supermarkets in October and is thought to have exploited legacy equipment exposed to the internet with default passwords.
A second wave in early November hit more than 100 shops across several European countries, including the Netherlands, Sweden, Ireland, Belgium, Finland, Estonia and Germany. The UK was affected, but only slightly. The NCSC confirmed a small number of incidents, mostly at meat suppliers, and called the national impact minimal. However, researchers have found more than 800 internet-exposed fridge interfaces tied to British retailers, so the risk remains.
What should I do?
Organisations should reassess their risk profiles rather than assume existing controls remain sufficient. This includes considering whether vulnerabilities are addressed quickly enough, whether suspicious activity can be detected promptly and whether essential services can be restored following disruption. Dependencies on suppliers and external infrastructure should be included in that assessment.
- Identify systems, information and suppliers that support essential business activities. Review internet facing technology and remove access that is not genuinely needed. Assess unsupported software, outstanding security updates and known exploited vulnerabilities, giving priority to weaknesses that could expose critical services.
- Strengthen foundational controls before relying on new AI-based security products. Apply relevant patches, implement multi-factor authentication, restrict privileged access and segment networks to limit the spread of an intrusion. Maintain protected backups and test that systems and data can be restored within acceptable timescales. For operational technology, coordinate updates and configuration changes with engineers and vendors to avoid creating safety or availability problems.
- Review incident-response and business-continuity arrangements against scenarios involving ransomware, extended outages and supplier failures. Exercises should involve senior management as well as technical teams, so that decisions about service restoration, communications and regulatory reporting can be made quickly. Ask when recovery was last tested successfully and whether the organisation could continue operating while a critical supplier was unavailable.
- Training for employees should address increasingly convincing phishing and impersonation attempts, including those assisted by AI. Sensitive requests, such as changes to payment details or account access, should be verified through a separate, trusted channel. Staff should also have a straightforward way to report suspicious activity promptly.
- Follow NCSC’s recommendation to treat Cyber Essentials as a starting point rather than a complete security strategy. Use relevant NCSC guidance, including the Cyber Assessment Framework and ransomware resources, to identify improvements proportionate to the organisation’s size, complexity and risk. Leaders should ask their security teams which significant weaknesses remain unresolved, what investment is needed and how progress towards greater resilience will be measured.