Cloudflare data reveals bots now generate 57.4% of all web traffic

A milestone has been reached in the history of the internet: for the first time, automated bots have overtaken humans as the primary source of web traffic.

Fresh data from Cloudflare, which handles traffic management for some of the largest websites on the planet, reveals that approximately 57.4% of all web activity recorded over the past seven days originated from automated systems rather than real users.

Bot traffic now dominates

Cloudflare's figures show bot activity fluctuating between 52% and 62% at any given point throughout the day, with human users accounting for roughly 42.5% of traffic.

The company had not anticipated AI-driven bot traffic would surpass human levels until next year, meaning the current data represents an earlier-than-expected arrival of a shift that the industry had long seen coming but had not yet fully prepared for.

AI agents are the cause, not traditional crawlers

The surge has not been brought about by the conventional search indexing bots deployed by the likes of Google, which have been a feature of the web since its earliest days.

Rather, the rise is being driven specifically by AI agents — systems that scrape content for model training, and agentic bots that carry out tasks autonomously on behalf of users of AI assistants and chatbots.

Where bot traffic is highest

The geographic breakdown of Cloudflare's data adds important context to the headline figures. Gibraltar tops the list, with 92.1% of its traffic attributed to bots. Singapore follows in second place at 76.3%, with Iran close behind at 76.2%. Ireland ranks fourth at 72.8%, and the Netherlands fifth at 68.8%.

The concentration of bot traffic in countries such as Singapore, Ireland, and the Netherlands is less a reflection of where AI activity originates and more a consequence of where data centres are located — these are the destinations to which automated traffic is being directed, not its source.

Why the numbers can be misleading

Cloudflare's methodology is based on page load counts, which means the metric captures every instance a page is accessed, regardless of what happens next.

When a human visits an article or watches a video, that registers as a single visit and reflects genuine engagement. A bot indexing the same page, by contrast, loads it and moves on instantly to its next target.

For agentic bots acting on behalf of human users, the distorting effect is even greater. A single query submitted to a chatbot can trigger dozens of separate page loads as the AI agent draws from multiple sources to compile a response.

The result is that bots have dramatically inflated raw visit counts whilst delivering comparatively little in terms of the sustained engagement that publishers and advertisers actually value.