AI and Bots Surpass Human Share of Internet Traffic

Cloudflare data shows automated traffic has overtaken human activity on the web for the first time, with bots and AI agents generating approximately 57.5 percent of HTTP requests compared to 42.5 percent from people. Cloudflare CEO Matthew Prince said the crossover occurred faster than expected, having previously predicted it would not happen until 2027. The shift complicates engagement metrics, telemetry, and the provenance of web data used to train machine learning models.

AI and Bots Surpass Human Share of Internet Traffic Cloudflare data shows automated traffic has overtaken humans on the web for the first time, with bots and AI agents generating about 57.5 percent of HTTP requests versus roughly 42.5 percent from people, according to figures reported by Tom's Hardware and the New York Post. Cloudflare CEO Matthew Prince said the crossover arrived faster than expected, noting he had predicted it would not happen until 2027. Coverage attributes the surge to AI agents that fan out across many sites per task, alongside search crawlers and service-to-service calls. For data and ML teams, a majority-bot internet complicates engagement metrics, telemetry, and the provenance of web data used to train and evaluate models. What happened Cloudflare data indicates that automated traffic has overtaken human activity on the web for the first time, with bots and AI agents generating roughly 57.5 percent of HTTP requests against about 42.5 percent from people. The split was reported by Tom's Hardware, citing Cloudflare's latest measurements, and by the New York Post. Cloudflare CEO Matthew Prince said the crossover happened faster than he had predicted, having earlier expected automated traffic to eclipse humans only in 2027. Why bot traffic is rising Coverage attributes the jump to the spread of AI agents that browse many websites per request, sometimes visiting large numbers of pages to complete a single task such as comparison shopping. That behavior compounds long-standing automated categories, including search-engine crawlers, data-scraping bots, and service-to-service API calls. Before the recent AI boom, automated requests were commonly estimated well below half of traffic. Why it matters for data teams A majority-automated internet complicates measurements that assume human sessions, such as engagement, click-through, and pageview telemetry. It also affects the composition of web corpora used for model pretraining and evaluation, where scraped and machine-generated content can dilute or contaminate datasets. Bot-detection, rate-limiting, and fingerprinting become more central to keeping analytics and training data clean. What to watch Useful signals include corroborating measurements from other large networks and CDNs, publication of Cloudflare's underlying methodology and classification thresholds, and vendor guidance on filtering automated traffic from analytics. Practitioners may also track how agent traffic is labeled and authenticated, since distinguishing human-initiated from machine-initiated requests is becoming a core operational and research problem. Caveat on sources The headline figures trace to Cloudflare's own data as relayed by trade and general-interest outlets, rather than a detailed public methodology paper. Exact percentages vary by measurement window and whether requests or page views are counted, so the precise split should be read as a snapshot rather than a fixed ratio. Scoring Rationale A first-ever crossover where bots and AI agents generate the majority of web traffic is a notable, widely reported milestone that directly affects telemetry, analytics, and the provenance of ML training data. It reflects an important measurement signal rather than a frontier model release or regulation, so it sits in the notable band. Practice interview problems based on real data 1,500+ SQL & Python problems across 15 industry datasets — the exact type of data you work with. Try 250 free problems /problems