GEO
From September 15, Cloudflare default-blocks mixed-use AI crawlers on ad-supported pages for new customers, new sites and every free-tier account. Nobody gets a notification. Here is what changes, who is affected, and the five-minute check to run before the date arrives.
On September 15, Cloudflare changes a default. From that date it blocks what it calls mixed-use AI crawlers on ad-supported pages for new customers, new sites, and every free-tier account. There is no banner and no email. If the September 15 Cloudflare change applies to your site, the engines that answer buyer questions (ChatGPT, Perplexity, Gemini, Claude) can quietly lose the ability to read it. This post explains the Cloudflare AI crawler block in plain English: what is changing, who is affected, and the five-minute check we think every site owner should run this week.
Two glosses before the detail. A crawler is the software an engine sends out to read web pages. Cloudflare is a service that sits between millions of websites and their visitors and decides which traffic gets through. When Cloudflare turns a crawler away, nothing else about your site matters to that engine. It never sees the page.
TechCrunch reported on 1 July 2026 that Cloudflare will default-block "mixed-use" AI crawlers on ad-supported pages from September 15. Mixed-use means a crawler doing more than one job: indexing pages for search results while also feeding AI training or AI agents. Cloudflare's definition explicitly includes Googlebot, because the crawler that puts you in Google Search is the same one that supplies Google's AI answers.
The same report describes Cloudflare's Pay Per Crawl programme evolving into Pay Per Use. Instead of AI companies paying a flat toll each time a crawler fetches a page, payment would track how the content is actually used. The direction of travel is clear: Cloudflare wants AI access to the web to be something site owners charge for, and blocking is the lever.
The part we keep coming back to is the delivery. This is a default, not a letter. Per the TechCrunch report, nobody gets a notification. A site that falls under the new setting does not receive an email or a dashboard alert. It just stops appearing in AI answers, and the owner finds out later, if at all.
Three groups from September 15, per the report:
The stated scope is ad-supported pages. Read that carefully alongside the third group. Plenty of small business sites carry no advertising and may fall outside the stated scope. But the free tier is exactly where small sites live, the definition of ad-supported is Cloudflare's to interpret, and defaults have a history of widening once the plumbing exists.
Existing paid customers with existing sites keep their current settings, as far as the reporting goes. One honest caveat: we are describing a policy as reported in July, before it takes effect. Cloudflare may refine the details between now and the date. Do not take a blog post's word for your own configuration, including ours. Check.
Here is what to do today. The first two steps need nothing beyond a browser.
Why we care this much: in our readiness score, crawler access is worth 20 to 25 percent of the total, depending on the engine. The weighting is explained in our methodology, but the logic fits in one sentence. Readiness asks whether the door is unlocked, and every other signal we score (extractable HTML, structured data, being named on the pages engines read) only matters once a crawler is through the door. For the full picture of what engines see once inside, we covered it in What AI search engines actually read on your site.
We are not going to tell you that blocking is always wrong, because it is not.
If your revenue comes from advertising against pageviews, an AI engine that reads your work and answers the user's question without sending you the visit is taking value without returning it. Publishers who block are making a rational trade, and Cloudflare's move towards Pay Per Use exists precisely because that trade is real. If that is your business, blocking, or charging, can be the right call.
Feeding is rational for the opposite business. If your revenue comes from people finding you and becoming customers (a consultancy, a clinic, a shop, a software product, a local trade), then an AI answer that names you is a referral, and a blocked crawler is a referral that can never happen. Visibility in AI answers starts with being readable.
Two questions settle most cases:
Ad-funded publishers mostly land on block. Businesses that sell things mostly land on feed. Both answers are fine. What is not fine is inheriting the answer from a default that changed on a date you never heard about.
September 15 is not a catastrophe. It is a settings change. But silent settings changes are exactly how sites vanish from AI answers with nobody ever deciding they should. The check above takes five minutes and settles the question for your site.
Start with the free crawler check to see who can reach you today. Then get the rest of the picture, the part that matters once the door is open. Run a free scan at whocanfindme.com. No signup. Takes about ten seconds.
When a buyer asks an AI one question, the engine quietly turns it into about ten. Seer Interactive measured 10.7 hidden sub-queries per prompt across 501 prompts on Gemini 3. Most sites answer the headline question and none of the ten. Here is how to find your gaps.
Lire la suiteThe same question, same engine, same day can produce different answers. That is the nature of AI search, not a bug. The 3 rules we use to avoid showing you fake statistics, and how to read a visibility score without fooling yourself.
Lire la suite