Google Cloud unveiled Fraud Defense at its Cloud Next conference, positioning it as the next generation of reCAPTCHA for the emerging “agentic web.” Autonomous AI agents that can browse, shop and transact introduce novel fraud vectors, forcing security teams to rethink risk models. Fraud Defense promises a unified trust platform that can verify humans, bots and AI agents in real time.

New tools include an agentic activity dashboard that classifies traffic using Web Bot Auth, SPIFEE and traditional signals, and a policy engine that lets operators block or allow requests based on risk scores, automation type and identity. A QR‑code challenge adds an AI‑resistant human verification step. Existing reCAPTCHA sites inherit Fraud Defense automatically, with unchanged keys and pricing.

Google’s fraud‑intelligence graph already shields roughly half of the Fortune 100 and 14 million domains, giving Fraud Defense a collective immunity many on‑prem solutions lack. In internal tests the platform cut account‑takeover incidents by 51%, while AI‑driven shopping assistants are projected to lift average order value by 25%. The service now offers invisible verification for most users, tightening security without adding friction.