AI bots are rapidly taking over the internet and already account for significant share of web traffic
Remember the viral assistant OpenClaw? Until recently, we discussed it as just another amusing neural network add-on, but today it has become a symbol of an era's end. The network we built for humans is transforming before our eyes into a closed ecosystem for autonomous bots.
A new activity report on bot activity on the internet, along with data from infrastructure company Akamai, shows that AI bots already form a significant share of web traffic. The study also reveals an arms race between bots using clever tactics to bypass site defenses and the web resources themselves.
New traffic emperors
According to Akamai data, we've entered a phase where every 31st visit to websites is made not by a human. While AI scrapers made up only a tiny fraction at the beginning of last year (1 in 200 visits), by February 2026 their activity had grown exponentially.
The problem isn't that there are more bots, but in how they behave. More than 13% of AI agents now completely ignore the universal robots.txt file – a sort of "gentleman's agreement" of the internet, which site owners use to request certain data not be indexed. The share of such "brazen" bots jumped by 400%, turning data collection into full-scale digital looting.
Mimicking humans
Against this backdrop of aggressive content theft, major sites like Condé Nast are declaring war on AI corporations. The reason is that modern bots have learned to scan the network in real time. Previously, they simply collected databases for training models, but now they extract current prices, schedules, and news to present them to users within their own interface, thus depriving content creators themselves of traffic.
Akamai CTO Robert Blumofe describes the situation as an "arms race." Bots no longer present themselves as programs – they disguise themselves as regular browsers, mimic human reading pauses and chaotic mouse movements. As a result, site protection hits real users, forcing them to endlessly solve captchas, while advanced algorithms break through filters unnoticed.
Toshit Panigrahi, CEO of TollBit, says:
This isn't just a legal dispute about copyright. A new type of visitor has emerged on the internet that consumes content but gives nothing in return – no ad clicks, no community participation.
Marketing for ghosts and prospects
The most interesting developments are happening on the business side. Instead of blocking bots, some companies are starting to engage in GEO (Generative Engine Optimization). This is a new discipline that has replaced classic SEO (search engine optimization). If companies previously optimized texts for Google's search bar, now the task is to become "visible" to OpenClaw or ChatGPT algorithms.
In 2026, it's becoming more important for brands to have an AI assistant recommend them in its response than to be on the first line of search results. This creates a paradoxical situation – publishers hate scrapers but are forced to adapt to them to avoid disappearing from neural networks' "memory."
The world of the "Dead Internet," where bots communicate with bots while humans remain only on the sidelines of this chain, has already arrived. Welcome to cyberpunk. Although some companies offer site owners to charge bots for each content access, there's no widespread adoption of this approach yet.
Perhaps we'll soon see the internet divided into two parts – an open "wild west" for bots and closed, authenticated communities for living people, where entry will only be possible through biometrics.