Cloudflare Launches Trusted Agent Delegation Coalition and New Agent Visibility Tools

Details

Cloudflare announces the Trusted Agent Delegation (TAD) coalition to develop open, interoperable standards for AI agent identity, reputation, and accountability.
The core argument is that AI agents act as intermediaries that obscure the relationship between users and sites, requiring new mechanisms to preserve security and fraud prevention capabilities.
New Cloudflare features allow site owners to view agent traffic in Security Analytics, set per-agent access policies, and link agents to specific users via Account Abuse Protection.
The ecosystem needs protocols like Web Bot Auth (WBA) and MOLE to handle agent identity and rate limiting without relying on human-like behavior or CAPTCHAs.
Existing solutions like human-in-the-loop (HITL) and tight access controls fail to resolve issues of traffic commingling and accountability for actions taken by agents.
The Intermediation Principle states that business relationships and security responsibilities between users and sites must be upheld regardless of intermediation by agents.

Key point