Cloudflare Radar Reveals 57% of Web Traffic Is Bot‑Driven and AI Agents Rely on Accessibility Tree

Cloudflare’s Radar data for the week of May 30–June 5 2026 shows that 57.2 % of HTTP requests to HTML content were automated bots, compared with 42.8 % from humans, a crossover that occurred over a year earlier than CEO Matthew Prince’s 2027 forecast. The surge is driven largely by AI agents that scrape pages for real‑world users, and these agents depend on the browser’s accessibility tree—a semantic, stripped‑down representation of the DOM that screen readers and AI models use. Microsoft’s Playwright MCP uses the accessibility tree exclusively, while OpenAI’s Computer‑Using Agent relies on screenshots, and a hybrid approach combines both. The accessibility tree exposes each node’s role, name, state, and description, providing the structured data AI agents need to act efficiently. WebAIM’s 2026 Million report shows a 95.9 % WCAG failure rate on home pages, up from 94.8 % the year before, with 56.1 detected errors per page and 1,437 elements per page—a 22.5 % increase in complexity. The most common failures—low‑contrast text (83.9 %), missing alt text (53.1 %), missing form labels (51 %), empty links (46.3 %), and empty buttons (30.6 %)—directly strip meaning from the accessibility tree and hinder AI agents. As AI agents become the majority of HTML traffic, the quality of the accessibility tree will determine how well they can understand and interact with sites. Web developers should therefore audit and fix WCAG failures to ensure AI agents can reliably parse the accessibility tree.