Cloudflare Is Blocking Google by Default on September 15
From September 15, Cloudflare's default settings change how Google's crawler is treated. Here's what actually changes and the box to untick before the deadline.
Now the plain-English version. Cloudflare is the company nearly every website hides behind so it loads fast and doesn't fall over when traffic spikes. Google is the place nearly everyone goes to when they want to find a website in the first place. However, one of the companies that quietly runs the web is about to start turning away the company that quietly directs people to the web. Complicated? Maybe, especially as it sounds like the internet declaring war on itself.
But it isn't, quite. The real version is narrower and more interesting than the tweet, and if you run a site you should understand which one is true before September 15 comes and goes.
What's Actually Happening
For years, showing up in Google was all-or-nothing. Let Googlebot in, and you got search traffic to your site. You also, more recently, got your content fed into Gemini and into the AI Overviews that answer the question (or search queries, if we’re being technical) at the top of the page. The thing is, while AI overviews add a bit of authority to your content, they sort of replaced Featured snippets in outlook, and don’t necessarily link to your website. The whole point of putting our [great] content is that you want the traffic behind it; you were exchanging valuable information for hordes of people coming on to your website, and then giving you leverage to do whatever you wanted with the numbers.
Essentially, we had one bot for these two jobs, and I should add that there was no specific way to say yes to the first of those tasks and no to the second.
Cloudflare's move splits that bot's behaviour into three buckets. Search, which indexes you so people can find you later. Agent, a bot doing a task for a person in real time. And Training, which hoovers up your words to teach a model. Site owners can now allow or block each one on its own.
Then comes the major headline. Starting September 15, if a crawler does more than one of those jobs, Cloudflare judges it by its strictest behaviour. Googlebot crawls for search and training in a single bot. So if a site blocks training, it blocks Googlebot; and drops out of Google search as collateral.
That's where "block Google by default" comes from. It's technically downstream of true. But three things shrink it fast.
It only fires on pages Cloudflare's classifier decides are showing ads, not your whole site.
It only becomes the default for new domains and free-tier accounts that never touched their settings — everyone else keeps what they had.
And anyone can opt out in their security dashboard before the deadline, in about the time it takes to make coffee.
If I were to reword this development more accurately, it would be a bit uglier and duller:
On September 15, some ad-supported pages on some sites will stop letting a Google bot in for AI training, and a few of them will accidentally fall out of Google search unless someone unticks a box..
How the Cloudflare-Google Deal Fell Apart
To understand why any site would risk its Google traffic, you have to look at what that traffic is now worth. TL;DR: it's a fraction of what it used to be.
For twenty-odd years the bargain was simple and mostly fair. You wrote something, Google indexed it, and when someone searched for it Google sent them to your page. You got visitors, you got ad revenue or subscribers or customers, and Google got a cut of the attention by being the road everyone drove on. Publishers grumbled about ranking changes and algorithm updates, but nobody seriously considered walking away, because that would have meant site suicide. Or the web version of the Investigation Discovery tv show “Disappeared”.
Then, of course, Google started answering the question itself. AI Overviews sit at the very top of the results page and summarise the answer, stitched together from the same pages that used to get the click. The searcher reads the summary, nods, and closes the tab. Meanwhile, your page supplied the answer and got a bucketload of nothing for it. People in the industry have started calling this "Google Zero"; to signal the point where a site can rank perfectly well and still watch its search traffic drain toward nothing.
The numbers back up the entire mood of it all. Cloudflare reported this summer that automated bots now generate more than half of all web requests which just means that the web is busier with machines reading pages than with people reading them. Some of those bots are Google indexing you for search. A growing share, on the other end, are AI crawlers pulling your words to train models or to answer questions somewhere you'll never see.
Which just brings us back to good ol’ you. You're doing the writing, putting in the hardwork and all that only to find that increasingly the audience you’re creating for is practically software!
Why Would Anyone Do This On Purpose?
We’ve long established that the traffic Google sends back keeps shrinking as AI Overviews answer the question on Google's page instead of yours.
If you thought the “Google Zero” term was funny, wait till you find that the people in the recipe-and-review corners of the internet have started using the word hostage to describe the situation. Another site owner said the Google traffic was so small by now, after Google had buried them in results, that blocking the whole thing barely stung. Naturally, some of them go looking for the exit out of this crap arrangement. So, by blocking the training crawler, this is exacty how a small publisher stops feeding the machine that's eating its traffic.
The catch, however, is that this is a solid case of where the machine is bolted to the front door. You can't easily starve the AI without risking the search visibility in the same move, frankly because Google built them INTO the same bot. That's the actual fight here, and it's less "Cloudflare vs Google" than "a publisher trying to unbundle two things Google deliberately sold as one."
What Can Site Managers Actually Do?
If your site's on Cloudflare, and it probably is, open the security settings and look at how Search, Agent, and Training are set. If you want to stay in Google search while keeping your content out of AI training, allow Search and block Training, and use Google's separate Google-Extended control in your robots file.
I must warn you though: doing this stops your content from training Google's models but it doesn't pull you out of AI Overviews, which run off the regular search index. There's no clean single switch for "you can index me but don't quote me." At least not yet….
The deadline itself matters less. September 15 is the day a default flips, but the real event is that blocking AI at the infrastructure layer just became a checkbox instead of a project. When a company that sits in front of a large chunk of the web makes "keep the AI crawlers out" the out-of-the-box setting for new sites, it changes what publishers assume they're agreeing to when they put content online. The old default was an open book. The new default, for a growing slice of the web, is closed until you decide otherwise.
Cloudflare has also floated the other half of this: a way for sites to charge crawlers for access rather than simply blocking them, so the words you write become something you can meter and bill for instead of something you just hope to see referral traffic from.
The fight isn't really about whether Googlebot can read your page. It's about whether the people who make the content the whole system runs on get to set terms, or keep taking whatever deal the platforms hand them.
For now, the terms you can set are small and specific.
FAQs
Is Cloudflare blocking Google on September 15, 2026?
Not all of Google. From September 15, Cloudflare's default settings block AI training and agent crawlers on ad-supported pages for new domains and free-tier accounts that haven't changed their settings. Because Googlebot crawls for both search and AI training in one bot, sites that block training can block Googlebot as a side effect — but Google search crawling stays allowed by default.
Will my site disappear from Google search?
Only if you block the Training category and haven't opted out, since that catches Googlebot too. Most existing paid accounts keep their current settings and aren't affected. You can check and change this in your Cloudflare security dashboard before the deadline.
Does this affect every website on Cloudflare?
No. The new default applies to new domains, newly added sites, and existing free-tier accounts that never touched their crawler settings. It only triggers on pages Cloudflare classifies as showing ads, not your whole site.
How do I keep Google search but block AI training?
Allow the Search category and block Training in Cloudflare, and add Google-Extended to your robots.txt to opt out of Google's AI training. Note this doesn't remove you from AI Overviews, which pull from the regular search index.
What is "Google Zero"?
The point where a site ranks well in Google but still loses search traffic, because AI Overviews answer the query on Google's own results page and the user never clicks through to the source.
