Cloudflare launches Trusted Agent Delegation so sites can recognize AI agents acting for users
On November 5, Cloudflare wrote that cloud-hosted personal AI agents sit between users and sites, so sites often see the agent instead of the person—making account sharing, friendly fraud, identity checks, and mixed traffic harder. It launched the Trusted Agent Delegation (TAD) coalition for operators, sites, payments, and identity vendors to align with IETF work such as Web Bot Auth, MOLE, and AI Preferences. Customers can now spot Personal Assistants in Security Analytics with per-agent paths and errors; Account Abuse Protection (early access) flags agent logins and lets sites act on a hashed user id without blocking every other user on that agent.
Why it matters
As agents browse for people, whether sites can see who acts—and who is accountable—will decide how open login, payments, and content stay to agent traffic.
Sources1
Written by dalao.ai: the title, summary and why it matters are our own words. The original reports are the record; their publishers keep authorship.