# Cloudflare, OpenAI pilot adds pages to ChatGPT Search

Published: 2026-10-20T15:05:00.464Z · Source: Cloudflare (https://blog.cloudflare.com/cloudflare-openai-discovery-agentic-web/)
Source date: 2026-10-15
Entities: openai, chatgpt, cloudflare

Cloudflare and OpenAI ran a research pilot on about 1% of Cloudflare's network, starting in July 2026, using Cloudflare signals such as page-update recency and traffic quality to help OpenAI's crawler find pages for ChatGPT Search, per a joint Cloudflare post co-authored with OpenAI staff on October 15. The result: "hundreds of millions" of previously undiscovered pages indexed, though neither company gave an exact count. The companies stress a signal does not grant crawl access: sites can still block OAI-SearchBot through robots.txt, and that permission is separate from whether a page's content can be used for model training.

Why it matters: Network-level signals about page freshness and traffic quality give OpenAI a way to find pages that ordinary crawling can miss, adding a new, largely invisible factor — how a site performs on Cloudflare's network — to whether ChatGPT Search finds it at all.

## What this answers

**Can a website stop OpenAI from using Cloudflare's signals to find its pages for ChatGPT Search?**

Yes — Cloudflare and OpenAI say a network signal about a page does not grant permission to crawl it, so a site can still keep OAI-SearchBot out entirely via robots.txt, separately from any model-training permission.


Canonical: https://anythingengineoptimization.com/item/2026-10-20-cloudflare-openai-pilot-adds-pages-to-chatgpt-search/
From Anything Engine Optimization (AEO Wire) — https://anythingengineoptimization.com/ · Standards: https://anythingengineoptimization.com/standards/