Cloudflare and OpenAI launched a research pilot in July 2026 to test whether signals from Cloudflare’s network can help ChatGPT users discover new content more efficiently.
The companies said early results were positive. The pilot brought previously unknown pages into OpenAI’s search index, giving ChatGPT more sources to cite in its responses.
Pilot used about 1% of Cloudflare’s participating queue
The test drew on about 1% of the participating queue across Cloudflare’s global network. OpenAI said that helped it discover and index hundreds of millions of new web pages.
OpenAI independently evaluates those pages and ranks them by relevance and usefulness. In ChatGPT Search, users can follow links back to the original content and its publisher.
Companies flagged several page-quality issues
According to the observations shared by the two companies, automated traffic can make pages that do not actually exist appear worth attention. They also said some sites redirect arbitrary addresses to the homepage, while other addresses lead to duplicate content.
The pilot is also being used to test signals that can identify potentially harmful pages, including fake login pages inserted into legitimate websites. Cloudflare and OpenAI said such labeling should target a single page rather than an entire domain.
Signals are not permission to access content
The same signals are also being used to filter out pages that do not exist or that redirect elsewhere, and to judge whether unchanged pages are worth revisiting.
The companies said page signals do not amount to permission to access content. The research uses web metadata for search and discovery, rather than using page content for model training.
Website owners can use Cloudflare’s AI Crawl Control to understand and manage how AI systems access their sites. In the next phase, the companies plan to improve page selection and test faster, more reliable change signals.