Cloudflare's new setting lets sites refuse AI training while keeping Google indexing

shashib · x · 2026-09-24

On September 15 Cloudflare launched a "Disallow AI Training" dashboard setting that lets sites refuse AI model crawlers while still allowing Google, Bing, and Apple search indexing, addressing the long-standing blur between search and training permissions in robots.txt.

The piece also warns of the opposite failure mode: over-blocking can make pages invisible to AI assistants, so a day's work on a press release may never reach a reader.

Original post →

More from Infra

Infra channel →