Cloudflare: AI agents are not humans; sites only see an intermediary

Cloudflare published a post titled “Agents aren’t humans (but they kinda are)” (November 5, 2026), which formulates the “intermediation principle” — a site sees not a human, but an AI agent acting on their behalf — and announces the Trusted Agent Delegation coalition based on Web Bot Auth, MOLE, x402 standards, and the IETF AIPREF working group.

Compiled by Sergey KostenchukPublished 2026-11-05Updated 2026-11-05

2026-11-05 Coding OpenAI

🤖 Cloudflare: AI agents are not humans; sites only see an intermediary

Cloudflare formulated the “intermediation principle”: when an AI agent acts on behalf of a user, the site loses capabilities for identification, reputation, and accountability. The company listed 14 problems — from account sharing to traffic commingling — and announced the Trusted Agent Delegation coalition based on Web Bot Auth, MOLE, x402 standards, and the IETF AIPREF working group.

🌍 Instead of blocking agentic traffic, Cloudflare proposes regulating it with open standards: the costs of agent errors fall on sites rather than operators, and TAD and IETF may become the basis for a de facto standard for agent identification.

👤 For those building agents or sites, this is a checklist: intermediation risks (friendly fraud, traffic commingling) and available Cloudflare per-agent access policies and Account Abuse Protection with Hashed User ID.

Source 1: https://blog.cloudflare.com/trusted-agent-delegation/

Sources