Google Is Catching AI Spam With Its Own AI. 40 Billion Pages a Day
Google just confirmed what I've been saying for a while: AI is now fighting AI spam. The scale — 40 billion spam pages a day. That's not a typo and not a figure of speech. That's a volume human moderation could never handle, which is exactly why Google had no choice but to set machine against machine.
I know this pattern too well from the affiliate niche. I've watched the same scheme play out over and over: take a template, generate a thousand pages around near-keyword queries, swap the brand name — and wait for traffic. Three years ago, this still worked. Not anymore. And it's not that the algorithm got "smarter" in some abstract sense. It's that it learned to do exactly what a good editor does: look at the page, not the domain, and ask one simple question — is there actually a point to this?
The four pillars everyone forgets
Google is direct about its quality criteria: effort, originality, skill, accuracy. Sounds obvious, but almost nobody actually checks their own pages against it. Take any article on your site and ask yourself honestly: if you swapped the operator's name for a competitor's, would anything about it actually change? If not — you don't have content, you have a wrapper. And wrappers are exactly what's getting cut down first right now.
The real problem isn't AI as a tool. AI generation by itself isn't a crime — otherwise half the industry would've collapsed at once. The problem is low-effort content with no editing, no added expertise, no human who actually knows the subject and put something of their own into it. Google has learned to tell the difference between text written just to exist and text written to actually answer a question.
Scale is no longer an advantage
The old logic was simple: more pages, more shots at catching long-tail queries. Now it works the other way. 100 genuinely good pages beat 10,000 templated ones. Scoring happens at the page level, not the domain level, and that kills the old trick where one "fat" authoritative site could carry thousands of junk URLs on its accumulated trust.
In iGaming and betting, this hits especially hard on doorway pages and spun content split by geo and language. A hundred versions of the same casino review for different countries with minimal edits is exactly what the filters are catching first now. And it's not about a formal penalty for a "spam technique" — it's that for the actual user, that page is useless, and Google is increasingly measuring exactly that.
What to actually do about it
In your shoes, I wouldn't start with panic — I'd start with an audit. Go through your site and honestly flag which pages exist only because they were easy to generate, not because they mean anything. Kill any cloaking or doorway schemes if you've still got them — that's not a risk anymore, it's a guaranteed loss. If you're using AI to write — use it as a draft, not a finished product: editing, personal experience, specific numbers, and fact-checking are mandatory.
And yes, this is a long game. Fast scale doesn't give you fast results anymore — it gives you a fast reversal. Patience and steady work on the quality of each individual page now beats any volume play. That's a fact, and it's easier to accept it now than later.
If you need a sober audit of your content — one that tells you what will actually survive the next wave of spam filters — that's exactly what we do: check our services.