Skip to main content

AI Brand Mentions: Is This Just 2010 SEO All Over Again?

Short answer: yes, a big part of today's AI brand mentions game is 2010 SEO wearing new clothes. In 2010, people wrote pages stuffed with one keyword that gave visitors nothing. In 2026, people write pages stuffed with one brand name, often their own, so ChatGPT and Google AI Overviews repeat it. Google did let this work for a while. But on May 15, 2026, Google said its spam policies now cover AI answers too. The window is closing, just like it did in 2011.

This is not a how-to guide. It is a mirror. On the left is what SEO looked like in 2010. On the right is what "AI visibility" looks like in 2026. Look at both, and you will see the same movie playing again.

old computer monitor representing 2010 SEO and keyword stuffing
Photo by ninjason on Unsplash

The Mirror: 2010 SEO vs 2026 AI Brand Mentions

2010 SEO trick2026 AI versionWho did it help?
Repeat the main keyword 20 times on a pageRepeat the brand name across dozens of "best X" pagesThe site owner, not the reader
Article directories full of thin posts with one keyword linkPaid or AI-written posts on random blogs, each "mentioning" the same brandThe brand, not the reader
Exact-match anchor text everywhereExact-match brand claims everywhere: "X is the best tool for Y"The brand, not the reader
Hidden text (white on white)Hidden instructions for AI bots inside HTML, and pages shown only to AI crawlersNobody honest
Fake forum and blog commentsFake Reddit threads and fake reviews to create "consensus"Nobody honest
Content farms with thousands of pagesThousands of AI-written "best X 2026" listicles with only the year changedNobody honest

The pattern is the same. In 2010, the goal was to make Google think a page was relevant. In 2026, the goal is to make AI think a brand is popular. In both cases, the visitor gets a page that was written for a machine, not for them.

What 2010 Content Actually Looked Like

If you started SEO before 2011, you remember this. A page about "cheap car insurance" would say "cheap car insurance" in the title, the first line, every heading, the image name, the footer and ten more times in the body. The text said nothing new. It was written so the keyword density looked "right".

Then came article directories. You wrote one thin article, spun it into 50 versions, and posted each with a keyword link. No reader ever learned anything. The only goal was the link and the keyword.

It worked. That is the uncomfortable part. Google's systems in 2010 could count words and links much better than they could judge honesty or usefulness.

What 2026 AI Brand Mention Content Looks Like

Now look at today's "AI visibility" playbook. The most common trick is the self-promotional listicle: a company publishes "Best [category] tools in 2026" on its own blog and ranks itself number one.

  • Writer Will Oremus counted 60+ self-ranking listicles from Shopify and nearly 300 from ClickUp. When he asked ChatGPT for the best way to set up an online store, it named Shopify first and cited Shopify's own ranking pages.
  • A Peec study of 232,000 AI citations (December 2025 to February 2026) found that about 11% of citations came from self-promotional listicles. ChatGPT filtered them best (about 4%), while Google AI Mode and Perplexity were around 10 to 11%.
  • In May 2025, SEOs showed that simply publishing an article naming a company "best in category" was enough for Google's AI Overviews to repeat the claim as if it came from an independent source.

Then there is the paid side: services selling "guaranteed ChatGPT mentions", AI-written posts on low-quality blogs, and fake Reddit threads where one account asks a question and another "happens" to recommend the brand.

Ask yourself one question about any of these pages: did a real reader learn anything? If the honest answer is no, it is the same 2010 content. Only the target changed, from a keyword to AI brand mentions.

AI chatbot screen showing how AI brand mentions are repeated in answers
Photo by boliviainteligente on Unsplash

So Why Did Google Accept It?

This is the real question. Here are four honest reasons.

1. AI checks if a claim exists, not if the source is honest

When an AI answer is "grounded", the system looks for pages that support a claim. If ten pages say "Brand X is the best", the claim looks supported. The system is weaker at asking who wrote those pages and why. That is exactly the gap keyword stuffing used in 2010: Google could count, but could not yet judge intent.

2. New systems always lag behind spammers

Keyword stuffing worked for years before Google's Panda update hit content farms in 2011. Link schemes worked until Penguin in 2012. AI answers are only a few years old. Spammers always move first, and the cleanup always comes later.

3. Big brands did it too

When Shopify, HubSpot and other well-known companies publish self-ranking lists, it is harder to call the format "spam". It looks normal. In 2010, big sites also had keyword-heavy footers and city pages. Scale made it look acceptable.

4. Some of it is easy to fake and hard to detect

Research from Cornell (not yet peer-reviewed) reported that just 13 words of planted text could insert a recommendation into 38 to 51% of AI research agent sessions. If that is true even partly, detection is a hard problem, not a lazy one.

Google's 2011 Moment Has Already Started

Here is the good news for honest SEOs. Google has started doing what it did after 2010.

DateWhat happened
May 15, 2026Google updated its spam policies to say they apply to generative AI responses in Search, including AI Overviews and AI Mode. Pages served only to AI crawlers are treated as cloaking.
May 2026In its guide on AI search, Google said seeking inauthentic mentions across the web "isn't as helpful as it might seem."
June 2026Google's June spam update was the first enforcement wave with AI answer manipulation in scope.
April to June 2026SEO analyst Lily Ray studied 100 B2B "best software" queries in AI Overviews and found Google appears to have separated what it cites from what it recommends. Self-ranking pages could be cited while a competitor got the recommendation.

Ray also reported that sites hit hardest shared one pattern: mass-produced "best X" listicles, self-promotional comparison pages and scaled "alternatives" pages, with some losing 30 to 50% of visibility in their blog folders. Google even started adding warnings to some "best" queries. When she searched for the best SEO experts, AI Overviews warned that the field is saturated with self-proclaimed experts.

Sound familiar? It is Panda all over again, just aimed at AI brand mentions instead of keywords. For a refresher on what Google actually rewards, read our guide on how search engine rankings work and why you should care.

keyboard close up for writing honest content instead of fake AI brand mentions
Photo by christianw on Unsplash

The Irony: Stuffing Does Not Even Work Well in AI

The research paper that named "Generative Engine Optimization" in 2024 tested old tricks inside AI answers. Keyword stuffing actually lowered a page's visibility in generated answers compared with doing nothing. Adding real statistics, quotes and sources helped more.

Lily Ray's data pointed the same way: the brands AI recommended had far more referring domains and far more mentions from other sites. Calling yourself the best changed little. Other people calling you the best changed a lot.

Fake Mentions vs Real Mentions: The Simple Test

Fake (2010-style)Real (lasts)
You rank yourself #1 on your own blogAn independent reviewer tests you and ranks you
Paid post on a random blog that exists only to sell linksGuest post on a real site in your niche that real readers visit
AI-written post that only repeats your brand nameArticle with real data, examples or experience, where your brand fits naturally
Fake Reddit question and fake answerReal customers talking about you because you helped them
Same "best X 2026" page with only the year changedA comparison page that is actually re-tested and updated

One rule decides which column you are in: would this page exist if AI and Google did not exist? If yes, it is real marketing. If no, it is 2010 SEO.

What Smart SEOs Should Do Now

  1. Stop scaling self-ranking listicles. If you keep one, be fair: test competitors honestly and say clearly that you are one of the options.
  2. Earn mentions on real sites. Guest posts still work when the site has real readers and the article teaches something. That is the difference between outreach and spam. See how SEO authority links boost trust, traffic and rankings for the same lesson on links.
  3. Put real data in your content. Numbers, screenshots, case studies and first-hand experience help both readers and AI answers.
  4. Never use hidden text or AI-only pages. Google now calls pages served only to AI crawlers cloaking.
  5. Do not sell "guaranteed AI mentions". Nobody controls AI answers. Promise better visibility through real work, not a fake guarantee.

If you sell outreach, know exactly what you are offering. Our guide on the difference between guest posting and link insertion helps you explain it honestly to clients.

SEO working on real content to earn honest AI brand mentions
Photo by christinhumephoto on Unsplash

FAQs

Are AI brand mentions just keyword stuffing with a new name?

The manipulative kind is. Pages written only to repeat a brand name for AI, with nothing useful for readers, follow the same logic as 2010 keyword stuffing. Real mentions earned through useful content are different and still valuable.

Does Google allow manipulating AI Overviews?

No. Since May 15, 2026, Google's spam policies say they apply to AI Overviews and AI Mode. Google has also warned that seeking inauthentic mentions is not as helpful as it seems.

Do self-promotional "best of" lists still work?

They worked well in 2025 and still get cited sometimes. But 2026 studies show Google may cite these pages while recommending a different brand, and many sites using them at scale lost visibility.

Is buying guest posts for AI brand mentions spam?

It depends on the site and the content. A useful article on a real site with real readers is marketing. Mass-produced posts on link-selling blogs, made only to plant a brand name, look like the same link schemes Google punished in 2012.

Will there be a "Panda update" for AI search?

It has partly started. The May 2026 policy change and the June 2026 spam update brought AI answer manipulation into Google's spam rules. More enforcement is likely as detection improves.

Conclusion

Yes, a lot of today's AI brand mentions work is 2010 SEO with a new target. Then it was a keyword; now it is a brand name. Then it was content farms; now it is AI-written listicles. Google accepted it for a while because new systems are easy to fool and slow to fix. But history is repeating fully, including the ending: Google has already started calling AI answer manipulation spam. The SEOs who survived Panda were the ones who wrote for people. The same will be true this time. If you want growth that lasts, start with the winning mix of SEO, content, backlinks and social media.

Leave a Reply