Well well well.. here we are at last, Google is having a content problem, so much content, so little compute. So now what do they do how does Google battle the content farms that are cranking content beyond measure? Well, of course, battle AI with your own AI. Last week in Barcelona, Gary Illyes said the quiet part out loud: "scaled content is now becoming more of a problem than link spam."
I can already see the LinkedIn posts. Half of them will say links don't matter anymore, because Google isn't even worried about them. The other half will say link builders can relax, because the heat moved somewhere else.
Both are wrong. Links didn't get less important. Faking them just stopped being the cheapest way to cheat.
What Google actually said
Google filters out 40 billion spam pages a day, and it's now using AI to catch AI slop. Illyes shared the numbers at the Search Central Live Deep Dive in Barcelona (September 30 to October 2), and Search Engine Roundtable wrote them up this morning.
The short version:
- The problem is volume. There's way more new content being published, so Google is shipping more spam updates.
- The tool is AI. Google is using AI to hunt down the low-effort, scaled stuff that AI made easy to produce.
- The ranking of problems flipped. In Illyes' words, scaled content "is now becoming more of a problem than link spam" (We Are Roast recap).
Add it up and you get a trend. As Lily Ray points out, Google has released four spam updates in 2026, compared to one in 2025. She also lines up a month of documentation changes:
- AI content now has to be fact-checked.
- Quality raters now judge content on effort, originality, talent and accuracy.
- Fake authors are now officially "deception."
Her call: a big update is coming. Google usually rewrites the rules right before it enforces them.
The shortcut moved
Spammers don't really love links. They love whatever's cheapest.
For about fifteen years, the cheapest way to look important to Google was to fake the votes: PBNs, link farms, comment spam, $40 guest posts on sites nobody reads. Google spent a decade making those more expensive and more likely to blow up in your face. So the cheaters moved on.
Then generative AI made it nearly free to publish ten thousand pages before lunch. Now the cheap shortcut is pretending to be the content, not faking the votes for it. Of course scaled content is Google's biggest headache. It's the new bargain bin.
Notice what didn't move. A real editor at a real publication deciding your work is worth citing was hard in 2012, and it's hard now. You can't scale it with a prompt. That's not a bug in link building. That's the whole point of it.
Paid mentions are link spam in a trench coat
The old shortcut didn't die, though. It got a new name.
If you've been on LinkedIn in the last six months, you've seen the pitch: "paid mentions" or "AI visibility placements" that will get your brand cited by ChatGPT. Pay a site, get a mention, and the robots will notice.
Lily Ray flags exactly this in her prediction post. If a paid placement passes ranking credit without a nofollow or sponsored tag, it breaks Google's link spam policy, no matter what the invoice calls it. Calling it GEO doesn't change what it is.
That's the part that worries me for buyers. The same people Google just spent a month warning about scaled content and fake authors are now selling "AI visibility." The packaging is new, but the footprint is the same old one.
Why the hard thing pays twice
An editorial link still pays in two currencies: rankings and AI citations. Scarcity is the reason both of them hold their value.
Google more or less confirmed the first currency in Barcelona. Illyes closed the event with "AI results are built on SEO infrastructure, SEO isn't dead." AI Overviews and AI Mode pull from the same system that ranks pages. If you don't rank, you don't get cited.
The second currency showed up in the same session. Alex Wright of iDHL shared a case study where 83% of a brand's AI citations came from sites the brand doesn't own (recap). There's a word for getting your name on sites you don't own. It's link building.
So here's the math as Google rolls out more spam updates:
- Scaled content gets cheaper to make and riskier to publish.
- Paid mentions get you the same risk with a GEO label on top.
- Editorial links and earned media stay hard to get, so they keep their value in both currencies.
If the big one hits
Nobody outside Google knows when the next major update lands, or whether it lands at all. But recovering from a bad core update typically takes 3 to 6 months, and sometimes up to a year, per Illyes' own timing ranges (Search Engine Journal). Do your audit now, not later.
Five questions to ask about your link profile this week:
- Would this link exist if you hadn't paid for it? If the answer is no and it isn't tagged sponsored, that's a risk, not an asset.
- Is the linking page itself scaled content? A link from a 40,000-page AI content farm is borrowing trouble.
- Did a real editor make a decision? That's what separates earned media, resource links and real partnerships from everything else.
- Is your vendor selling "AI visibility" with no editorial process? Ask how a placement gets approved. If nobody can name a human, you have your answer.
- Would you be comfortable showing the link to a Google quality rater? You're rating against effort, originality, talent and accuracy now. So are they.
The bottom line
Google didn't say links matter less. It said the cheaters found something cheaper.
That's good news if you've been doing this the slow way. Each spam update clears out some of the competition, and each AI citation study keeps pointing back to the same place: sites you don't own, written by people who chose to mention you. The shortcut moved, but the hard thing didn't, and the hard thing still pays twice.
