Cloudflare has proposed a collaborative framework to help web servers identify, permit, and manage traffic generated by artificial intelligence agents operating on behalf of users.
In an outline published by Cloudflare's Bryan Becker, the company highlighted a growing problem for web infrastructure: while AI agents perform tasks for millions of individuals, servers often encounter them only as intermediaries that mask the identity and intent of the actual user. This lack of direct visibility makes it difficult for website operators to distinguish legitimate, user-directed agent requests from unwanted automated activity or malicious bots.
To address this tension, Cloudflare is exploring a cooperative model designed to help systems verify and allow authorized AI agents safely. Concrete technical specifications, implementation standards, and deployment timelines for the proposed approach have not yet been announced.
