Cloudflare and Tools for Humanity propose Trusted Agent Delegation to tie AI agents to accountable users

Cloudflare and Tools for Humanity have proposed Trusted Agent Delegation, a framework meant to let websites recognise and hold accountable the people behind AI agents, Cloudflare announced on November 5. When many users act through one hosted agent, a site can no longer tell them apart, so it either blocks the agent for everyone or absorbs the costs of fraud and abuse.

The framework rests on identity, reputation and accountability. Sites could require one of four tiers of user identification depending on risk, from none, through a session or a stable account identifier, to a "unique" third-party identifier such as a government ID or Tools for Humanity's World ID for high-risk actions. When a site reports abusive users, the agent's operator would have to act against them, much as email providers act on spam reports; Cloudflare plans to require evidence of this before granting verified agent or operator status.

Cloudflare has launched a coalition site and says a working group of agent builders, KYC and payment providers is being formed, Biometric Update reports. Cloudflare's Bryan Becker said more than half of internet traffic is already bot-driven and he expects that to reach 99% "pretty quickly".