Hi everyone, this is Neo.

Today I want to talk about something that excites me and worries me at the same time: the full-platform backlash against AI slop has finally arrived.

Let’s start with the news.

Spotify officially announced it removed more than 75 million “spammy tracks” over the past year — bulk uploads, duplicate songs, and other low-effort content — while rolling out a whole suite of new AI policies: a spam filter, stricter impersonation rules, and a DDEX credit standard that requires creators to disclose AI involvement. According to Music Business Worldwide, Spotify now receives around 100,000 new song uploads every day, and a growing share of them are batch-produced with AI tools. Rival Deezer’s numbers are even wilder: 30,000 fully AI-generated tracks arrive every day, up from 10,000 in January and 20,000 in April.

Around the same time, LinkedIn tightened its detection systems against AI slop. The New York Times put it bluntly: platforms are no longer willing to carry the cost of policing junk created by people who treat generative AI as a volume machine.

A few days earlier, Wired’s Reece Rogers reported that communities like Reddit and Stack Overflow had introduced rules to limit AI-generated answers — because engagement collapses when slop spreads unchecked.

Put these stories together and the signal is unmistakable: platform tolerance for mass-produced AI content is falling off a cliff.

And if you run an independent e-commerce site, this is exactly the moment to stop and think.

Don’t celebrate yet — this hits independent sites harder than you think

I know what some of you are thinking: “Spotify is cleaning up music, LinkedIn is policing posts — what does that have to do with my little export e-commerce site?”

Everything. Because AI slop never started with music or social media. It started with content websites — our battlefield.

Think about the last 18 months. How many “AI bulk article generation” tools have flooded the market? How many outsourcing agencies promise “500 SEO articles a month”? How many sellers actually bought in and let AI “produce content” for them around the clock?

That playbook worked for a while, mostly because Google’s enforcement was slow. But here’s the key fact: Google never said “AI content is bad.” It said content must be helpful, reliable, and people-first. Google isn’t anti-AI. It’s anti-junk.

And now every platform is adopting that same logic.

Kevin Indig — former growth lead at Shopify and a well-known voice in the GEO space — recently coined a concept called “slop antibodies”: companies need internal systems that filter out low-grade AI output before it ever reaches users. His core argument: the problem is not the tools. It’s the people who treat them as production engines instead of editorial assistants.

I couldn’t agree more.

History repeats itself: yesterday’s content farms are today’s AI slop

Greg Jarboe, who wrote the SEJ piece that sparked this, drew a historical parallel I find spot-on.

Fifteen years ago, content mills pumped out human-written webspam at industrial scale, and it drove Google crazy. What did Google do? It didn’t “ban humans.” It upgraded its systems to reward quality — and the content mills died while people doing real work survived.

Today it’s the same story. Google’s official guidance has said it plainly for a long time: the problem isn’t whether a machine wrote the words. It’s whether the content is helpful, reliable, and high quality.

Jarboe also shared a personal story: years ago, his team used one tool to optimize and distribute press releases for an online financial newsletter. One release about under-the-radar micro-cap stocks pulled in a wave of new subscribers. Another one celebrating the newsletter’s anniversary pulled in exactly zero. Same client. Same agency. Same tool. The only difference was the content itself — readers care about what’s valuable, period.

Here’s the line he said that I want to put in bold: “Tools amplify what you feed them. They do not fix it.”

Neo’s take: this backlash is the first real reckoning for the “AI gold rush”

Looking at this through the eyes of an independent site owner, I see three layers.

Layer one: the arbitrage window is closing. The AI content arbitrage play — cheaply producing content at scale to farm search traffic — has a much shorter shelf life than most people assume. The moment every seller knows the trick and every tool supports it, it stops being arbitrage and becomes a race to the bottom — and platform enforcement just ends that race early. Building an AI content farm in 2026 is like opening a content mill in 2012: using an old map to find a new continent.

Layer two: Google’s AI search makes junk even more visible — and even more vulnerable. AI Overviews and AI Mode pull answers straight from the regular index. When the index fills up with junk, AI answer quality drops, and Google’s only response is to clean the index harder. In other words, AI search doesn’t make the volume play more effective. It makes the quality play more valuable. This matches what I’ve been saying all along: in the AI era, SEO is an arms race of content quality.

Layer three: this is good news for the rest of us. Platform crackdowns wash out the shortcut-takers. If you’re a site owner who actually puts care into your product, don’t panic — enjoy the cleanup. Fewer competitors, a cleaner index, and the content you worked hard on becomes easier to see and easier to cite.

Five practical rules: use AI to level up, not to produce junk

So how do you actually use AI without making slop? Combining the SEJ piece with what the best teams in the industry are doing, here are five rules you can apply today.

1. Let AI analyze. Don’t let AI write. Use AI to cluster keywords, surface content gaps, and digest primary sources — then let a human decide what matters and how to say it. AI is brilliant at organizing information neatly. It’s terrible at knowing what actually matters to your readers. That judgment is the editor’s job, and it’s not going anywhere.

2. Verify everything. AI is confident even when it’s wrong. If you publish its invented stats, quotes, or case studies without checking, you get burned twice: by readers and by the platform. Build a habit: every key number and every quote needs to trace back to a real, named source. Google rewards content grounded in real evidence.

3. Structure your pages for AI search. Clear explanations, authoritative citations, and direct answers — these help in Google rankings and they also make your content far more likely to be picked up by AI Overviews and ChatGPT. Don’t write your pages like a maze. If AI can’t extract your core point, neither can your customers.

4. Publish less. Edit more. This is the rule I want to shout from the rooftops. Volume is not a strategy. Rigor is. Instead of having AI pump out ten articles nobody reads, spend a week polishing one page that actually solves a real problem. The data doesn’t lie: carefully made pages outperform AI-batch pages over time — and the gap is widening.

5. Add signals AI can’t fake. Named experts, first-hand interviews, proprietary data, and real-world war stories — these are things batch AI content can’t produce. Even something as simple as “I hit this wall, spent X dollars, and failed Y times” is a trust signal that search engines and AI systems find genuinely hard to fabricate.

Wrapping up

The platforms have drawn the line: they’re not rejecting AI. They’re rejecting slop.

Treat AI as a shortcut and you’ll find yourself on the wrong side of that line. Treat AI as an assistant and you’ll find new ways to build something better.

As Emerson put it: if you build a genuinely better mousetrap, the world will beat a path to your door.

For independent site owners, the best strategy in 2026 isn’t who uses AI the hardest. It’s who has the best judgment.

The more powerful the tools get, the more valuable your judgment becomes. That goes for all of us.

I’m Neo. See you in the next one.