Cloudflare and Tools for Humanity, the company behind the World ID proof-of-human system, have proposed a framework called Trusted Agent Delegation (TAD) to help websites tell legitimate AI agent activity from abuse, according to a report by Biometric Update dated 8 November 2026. Cloudflare said it published the proposal on 5 November.
The partners argue that websites can often detect an AI agent but not the person behind it, what that person authorised, or who is accountable if something goes wrong. Under the proposal, sites could require one of four levels of user identification depending on risk: none, a single-session identifier, a stable account identifier, or a unique persistent identifier for high-risk actions, such as a government ID or a proof-of-human credential like World ID.
The framework also envisages reputation signals and an accountability rule under which agent operators would have to act against abusive users reported by a site, rather than the site blocking the agent entirely. A coalition website has launched, and a working group of agent builders, identity-verification firms and payment providers is reportedly forming.
It is a proposal rather than a finished standard, and no timelines were given for standards work. Biometric Update noted that persistent identifiers raise privacy and linkability concerns, which the proposal does not fully resolve.
Reporting note: This account relies on one trade outlet, Biometric Update, which draws heavily on statements by the proposers. Claims about the weaknesses of existing tools and about bot traffic come from Cloudflare and Tools for Humanity and were not independently verified.
Photo: Ana Las Heras / CC BY-SA 4.0 / Wikimedia Commons
