Websites increasingly receive requests from software acting on behalf of users. In a November 4 announcement, Cloudflare described tools to identify and control traffic associated with personal AI assistants.
The company describes visibility in Security Analytics and options for managing supported agents. It distinguishes this traffic from policies for search indexing and AI training crawlers. That separation matters: an owner may want to handle a user-directed assistant differently from a crawler collecting large quantities of content.
What controls can and cannot tell you
Agent identification can help an operator understand traffic and apply policies. It does not establish that every identified request is benign, or that every automated client will be recognized correctly. The announcement describes additional account abuse protection as early access; availability and plan requirements should be checked before relying on a feature.
Our take: web operators need clear distinctions between indexing, training and user-directed tasks, as well as a way to troubleshoot false positives. We have not enabled or tested these settings for either site.
Watch: Cloudflare on AI agents
This April 2026 Cloudflare discussion gives broader background on personal agents and security; it does not describe the November announcement.
Independent reporting based on the linked sources. Product and performance claims are attributed to their publishers unless noted.
