NCSC warns AI could accelerate attacks despite fall in serious breaches

The UK’s National Cyber Security Centre has warned of a potentially unprecedented rise in cyber incidents as AI helps attackers identify and exploit longstanding weaknesses faster. Chief executive Richard Horne said years of underinvestment have left organisations exposed, despite a decline in the most serious incidents. The agency received 1,831 attack reports, up from 1,727 a year earlier, while nationally significant incidents fell to 80 from 204. It cautioned against treating that decline as a trend.

Attackers still largely rely on established methods, including phishing, compromised accounts and vulnerable internet-facing systems. The NCSC has seen AI tools used in incidents, but not to enable new attack types. It expects AI to uncover numerous previously unknown software flaws, requiring organisations to prepare for a “vulnerability patch wave”. The agency judges it highly likely that by 2028, hostile states will routinely use AI against the UK and AI-powered malware will feature in a ransomware attack on critical national infrastructure.

For insurers, vulnerabilities in widely used software could create simultaneous losses across many policyholders. Unapproved employee use of AI tools adds data exposure risks, while state-backed operations complicate coverage exclusions. Horne urged organisations to disconnect technology that does not need internet access and treat Cyber Essentials as a baseline, adding protection according to their size, complexity and risk.