Key takeaway: Search Engine Journal reported that Cloudflare and OpenAI shared early results from a research pilot on October 15.
Search Engine Journal reported that Cloudflare and OpenAI shared early results from a research pilot on October 15. The companies said the pilot involved a cohort representing about 1% of Cloudflare’s total network and helped OpenAI discover and index hundreds of millions of new web pages. For SEO and search visibility, the development could make more pages discoverable by ChatGPT Search, but it does not replace the access rules that determine whether a crawler can reach a site.
How network signals affect discovery
Cloudflare’s network signals help OpenAI’s search crawler identify pages that may be useful, including pages with limited inbound linking. The examples cited in the report include a washing machine user guide, current passport requirements and fees, and a restaurant’s menu and opening hours. The signals can include information about when a page was updated, the quality of traffic reaching a site, and what changes were actually made to the page.
The signals are also intended to reduce unnecessary crawling. They can help identify requests for files that do not exist, URLs that repeatedly redirect to a homepage, and addresses that lead to duplicate content. Change signals can indicate whether a page should be revisited or whether another crawl is unlikely to add value. Cloudflare and OpenAI are also testing signals for potentially harmful pages, such as a fake login form placed on one page of an otherwise legitimate site. Early results suggest the approach may help identify pages to keep out of an index, although the report gives no figures for crawl-volume or server-load reductions.
Access still depends on crawler settings
The companies state that a signal about a page does not grant permission to access it. Website crawler controls still apply. A site that blocks OAI-SearchBot will not be included in ChatGPT’s search engine results, although it may still appear as a navigational link. Allowing OAI-SearchBot while blocking GPTBot can keep a site available to ChatGPT Search while indicating that the crawled content should not be used for training.
OpenAI independently evaluates and ranks the pages it indexes. Discovery and indexing therefore do not guarantee a citation, a prominent position, or referral traffic. The report also notes that tracking ChatGPT visits generally depends on UTM parameters or a referrer, and that some ChatGPT links previously lacked both.
A separate privacy change in ChatGPT
TechCrunch reported on October 20 that OpenAI had started rolling out Locked Chats. Users can protect a conversation with a four-digit PIN, or with Face ID or Touch ID on supported devices. Locked conversations move to a dedicated section and are separated from regular history, pinned conversations, and chat search results. Users can return to them later.
The feature is available across ChatGPT Free, Go, Plus, Pro, Business, Enterprise, and Edu plans, but not ChatGPT for Healthcare. The rollout is gradual, so it may not yet appear for every user. Locking a chat does not create end-to-end encryption, change retention periods, erase information already stored in Memory, or change the “Improve the model for everyone” setting. Temporary, archived, group, shared-link, scheduled-task, and meeting-recording chats are excluded.
Method and Limitations
The Cloudflare and OpenAI figures describe early pilot results. The report does not identify participating websites or say how many newly discovered pages were cited by ChatGPT or generated visits. The companies are continuing work on faster content-change signals and on measuring business outcomes, but no specific date for the next stage was reported. Website operators can currently inspect server logs and Cloudflare AI Crawl Control, then decide how OAI-SearchBot and GPTBot should access their properties.