Cloudflare and OpenAI published early results on October 15 from a search-discovery research pilot launched in July. The partners say signals from Cloudflare's network helped OpenAI find pages it had not previously indexed, giving ChatGPT Search more potential sources.
The participating cohort covered about 1% of Cloudflare's total network, according to the joint account. The partners say the pilot enabled OpenAI to discover and index hundreds of millions of new pages. They do not present an independent audit or a guarantee that any given site will be included.
Cloudflare provides discovery signals, while OpenAI independently evaluates and ranks pages for relevance and usefulness. Finding and indexing a page is therefore distinct from selecting it for a particular answer or sending a reader to its publisher.
The work also looks at avoiding wasted requests to nonexistent pages, broad homepage redirects and duplicate content. The partners say change signals can help identify pages worth revisiting rather than repeatedly downloading unchanged material.
The research uses web metadata for search and discovery rather than sharing page content for model training, the post says. It also states that a signal about a page does not grant permission to access it. Domain owners' access choices remain relevant.
What to watch: the next phase will test page selection, change signals and outcomes after discovery. The partners explicitly say they are still studying how discovery translates into business results. The announcement does not establish a measured referral lift or revenue gain for every participating publisher.