0
0
Artificial intelligence bots are now consuming a growing share of internet traffic, and new data suggests human visitors are steadily disappearing from publisher websites.
Traffic monitoring firm Tollbit reports that AI bot activity surged sharply throughout 2025, driven mainly by the use of AI tools as alternatives to traditional web search rather than by model training scrapes.
By the fourth quarter of 2025, there was approximately one AI bot visit for every 31 human visits to a website. This marks a major jump from the first quarter of the same year, when the ratio stood at one bot visit for every 200 human visits.
Tollbit notes that these figures may understate the real scale of AI activity, as modern bots increasingly mimic human browsing behavior, making detection difficult.
Human Web Traffic Is Declining
Alongside the rise in AI bot activity, overall human web traffic is shrinking. Tollbit data shows that between the third and fourth quarters of 2025, human visits to websites dropped by five percent.
This trend suggests a structural shift in how users access information online, with more people relying on AI tools to retrieve and summarize content rather than visiting original sources directly.
RAG Bots Now Drive Most AI Web Activity
Earlier concerns around AI web scraping focused largely on data collection for model training. While training scrapes still occur, they are no longer the primary source of AI traffic.
According to Tollbit, training related scraping fell by 15 percent between the second and fourth quarters of 2025. In contrast, retrieval augmented generation bots increased by 33 percent during the same period.
RAG bots are used by platforms such as ChatGPT and Google Gemini to fetch real time information from the web in response to user queries. Supporting this growth, AI search indexer traffic rose by 59 percent over the same timeframe.
Among major players, OpenAI accounted for the highest volume of scraping. Tollbit reports that ChatGPT User averaged five times more page requests than the next largest scraper, which was linked to Meta.
Users Are Starting Searches Inside AI Tools
User behavior data reinforces this shift. Marketing firm Eight Oh Two found that 37 percent of active AI users now begin their searches directly within AI platforms instead of using traditional search engines.
Pew Research also reports that 62 percent of adults in the United States use AI tools several times a week, indicating widespread consumer adoption even outside professional environments.
Most Scraped Website Categories
Tollbit identified business focused and professional websites, national news outlets, and lifestyle platforms as the most frequently scraped by AI bots.
However, the fastest growth in scraping occurred in the technology and consumer electronics sector, which saw a 107 percent increase since mid 2025. Business and professional sites followed with a 62 percent increase.
Tollbit attributes this rise to a growing number of user prompts in consumer AI applications focused on information retrieval.
AI Referrals to Publishers Continue to Fall
Despite heavy scraping, AI platforms send very little traffic back to original websites. Tollbit reports that referral clickthrough rates from AI apps dropped from 0.8 percent in the second quarter of 2025 to just 0.27 percent by the fourth quarter.
Even publishers with AI licensing agreements saw referral rates fall sharply to 1.33 percent, representing a significant decline over the same period.
This trend weakens traditional publisher revenue models that rely on organic traffic and ad impressions.
Internet May Become Bot First
Tollbit leadership warns that the current trajectory points toward an internet primarily consumed by AI rather than humans.
Based on current growth rates, AI bots could become the dominant visitors to publisher websites as early as this year. The shift appears difficult to reverse as users favor AI tools for speed, convenience, and research depth.
Long Term Risks for Users
Some studies suggest that heavy reliance on AI tools may negatively affect critical thinking, learning retention, and independent research skills. Research on students shows weaker knowledge retention among those who depend on AI for writing tasks compared to those who work without assistance.
Beyond learning concerns, AI adds another layer of content filtering on top of search algorithms, shaping what information users see and how it is presented.
