AI bot attacks rise 10x in a year, with UK third-most targeted nation

TL;DR: Thales’ 2026 Bad Bot Report records a 10-fold rise in AI-enabled daily bot attacks — from 2 million to 25 million in a single year. Bots now account for around 53% of all web traffic, with 40% classed as malicious. The UK is the third-most targeted country, behind the US and Australia, ahead of France. Concrete numbers UK CISOs and SME owners can take to a budget conversation.

The report frames the shift as broader than attack volume. AI and automation, Thales argues, have moved from something organisations tried to block to something they must also manage — including legitimate AI agents that look enough like bad ones to require operational handling rather than blunt blocking.

What the numbers actually show

Bot share of web traffic ticked up from 51% in 2024 to 53% in 2025; bad-bot traffic is around 40%. Industries hit include retail, financial services, education and government. Tim Chang, a Thales general manager for applications and security, told The Independent the operational challenge is no longer identifying that traffic is automated — it is understanding what each bot is actually doing, whether the activity aligns with business intent, and how it interacts with critical systems.

For UK readers, the third-place ranking is the more uncomfortable detail. Australia’s second place reflects a smaller online economy with disproportionately exposed surface area; the UK’s position alongside it points to attacker focus on English-language financial-services and retail estates with sophisticated payment rails worth probing.

Looking forward

For UK SMEs, the practical takeaway is that bot management is shifting from a single security control to a cross-functional concern. Detection alone — distinguishing human from bot — was the 2023 problem. The 2026 problem is intent inference: a bot scraping pricing data, an agent making a legitimate booking, an attacker probing for credential reuse and an LLM-driven research bot all look similar at the network layer. Investment priorities for UK firms running customer-facing web estates: behavioural rather than signature-based detection, agent identity standards, and clear policy on which AI agents are welcome on which endpoints. Treating “AI agent” as a third category alongside human and bot — rather than a subset of one — is now the more honest baseline.