Bots Take the Wheel: AI Agents Now Generate the Majority of Web Traffic

The balance of who—or what—actually uses the web is tipping toward machines. Recent figures from infrastructure providers like Cloudflare, combined with HUMAN Security’s 2026 State of AI Traffic & Cyberthreat Benchmark, suggest that automated bots and AI agents now account for a majority of HTTP requests on major networks. Cloudflare reports that roughly 57% of requests across its platform are generated by non‑human actors, leaving human users in the minority. HUMAN’s benchmark, based on more than one quadrillion observed interactions, finds that traffic from “agentic” AI—systems that browse, click, and submit forms autonomously—has grown nearly 8,000% year‑over‑year, with bots from large AI vendors such as OpenAI representing a substantial share. Traditional web crawlers and uptime monitors have been joined by sophisticated LLM‑driven scrapers and multi‑step agents whose behavior can be hard to distinguish from that of real users. For publishers and platforms, this shift raises uncomfortable questions about the reliability of ad metrics, the strain on infrastructure, and how to enforce terms of use or copyright when most visits are automated. It is also forcing security teams to rethink bot detection and rate‑limiting, as some malicious agents increasingly mimic human browsing patterns. Policymakers are only beginning to grapple with the idea that much of the “public internet” is now machine‑to‑machine, not human‑to‑website, communication.