For much of the internet’s life, you could assume a person sat at the other end of most clicks. That assumption has quietly flipped.
In June 2026, Cloudflare CEO Matthew Prince announced that on his company’s network, automated software now sends more traffic than people do. He sounded surprised: “Welp, that happened faster than I predicted. Thought it would be end of 2027, then early 2027, but agentic traffic growing so fast that bots have now passed human traffic online for the first time in the Internet’s history.”
This is one very large hosting company’s sample, not a count of the whole internet. But it is a large sample, and other measurements point the same way.
What the crossover figure actually measures
At the time of writing, Cloudflare’s own dashboard put the split at about 57.3% bots versus 42.7% humans for requests to load web pages. NBC News, reading the same figure at a different moment, logged 57.4% versus 42.6%. The numbers shift because the data does.
What the figure counts matters as much as the number itself. It measures requests for web pages, not hours spent or video watched. By those measures, people still dominate. Once you fold in apps, streaming, maps, and social scrolling, humans account for far more of the total time spent online. This is a crossover in one specific metric, not proof that the internet is now mostly robots by every count.
The method deserves caution too. Filippo Menczer, a professor of informatics and computer science at Indiana University, warned that “you can try to estimate the amount of bot traffic by looking at the agent strings, but these are very noisy estimates.” The overall trend is clear. The exact share on any given day is not.
Why the timeline moved forward
The accelerant is agentic AI: software agents that carry out a task on your behalf and visit far more pages than a person ever would. The gap is easy to picture. A human shopping for a product might open around five websites, while an agent can hit 5,000. Multiply that across millions of tasks and the mix tips fast.
The growth is steep. HUMAN Security reported that automated traffic grew roughly eight times faster than human traffic across 2025. Prince’s earlier forecast of a late-2027 crossover arrived roughly eighteen months early.
Not all bots are the same
Let’s clear one thing up; a bot-majority web is not automatically a hostile one. The category runs from search-engine crawlers that have indexed the web for decades, to AI agents booking travel on someone’s behalf, to hostile traffic that hunts for weaknesses. Imperva’s 2026 report found bad bots made up 53% of internet traffic in, compared to 37% in their 2025 report.
The line between helpful and harmful automation is thinner than it looks. In HUMAN Security’s announcement, Lindsay Kaye warned that trusting agentic AI without scrutiny could lead to stolen passwords, misused data, and unwanted outcomes when shopping.
A web built for readers that no longer has them
The deeper issue is how the web is built. Stu Solomon, the CEO of HUMAN Security, put it this way: “The internet as a whole was created with this very basic notion that there’s a human being on the other side of the computer screen, and that notion is very rapidly being replaced.” Solomon runs a company with a stake in the answer, but the observation echoes across the reporting.
Systems built on that human assumption are already straining. Login pages are a clear example: automated attempts to break into accounts have climbed as attackers point bots at sign-in screens. Site owners are pushing back on the crawling too. Cloudflare says it has blocked over 416 billion AI bot requests since July 2025 at site owners’ request.
Where this settles is unclear. Solomon argues against treating all automation as a threat: “This notion of machine bad, human good just is not realistic. You have to live in a world where machines are acting on our behalf, and we have to establish a level of trust that’s persistent over time.” That prescription happens to describe the market his company operates in, which is worth keeping in view. The design question underneath it is real, and it now sits in front of everyone who builds for the web.
The pages, forms, login screens, and content across the internet were shaped for a reader with eyes and patience. On at least one large network, most of the visitors now reaching for those pages have neither, and the pages were not designed with them in mind.
