I’ve been doing this long enough to remember exactly where I was when Penguin hit in 2012. I had clients who’d spent two years building link profiles through directories, sitewide footer links, and paid guest posts on sites nobody actually read. Overnight, rankings that had taken years to build evaporated. Some of those businesses never fully recovered. Others rebuilt from scratch, this time the slow, boring, correct way.
I bring this up now because I keep having the same conversation with new clients, over and over, except the word “link” has been quietly swapped out for “citation” or “mention” or “AI visibility.” The tactics being pitched to nervous business owners right now — buy placements on sites that will “get you cited by ChatGPT,” pay for bulk mentions across low-quality directories to “train the AI to trust you,” stuff branded phrases into forum comments hoping a model scrapes it later — these are not new ideas. I watched almost this exact playbook get demolished by Google over a decade ago. And I don’t think the AI platforms are going to be any more forgiving.
So let’s talk honestly about whether we’re setting ourselves up to relearn the Penguin lesson, just with a different scoreboard.
A Quick Refresher on What Penguin Actually Punished
Before Penguin, link building had become a numbers game. If you wanted to rank, you got links — as many as possible, from wherever possible, with exact-match anchor text stuffed in for good measure. Nobody was asking whether a link made editorial sense. They were asking whether it counted.
Google’s response wasn’t subtle. Penguin specifically targeted manipulative link schemes: link farms, paid links disguised as organic mentions, over-optimized anchor text, and sitewide links that existed purely to pass authority rather than to help an actual reader find something useful. The algorithm didn’t just devalue these links — in many cases it penalized the sites that had built them, sometimes severely.
The deeper lesson, the one a lot of people missed at the time, wasn’t really about links at all. It was about intent. Google was drawing a hard line between signals that occurred naturally because content had genuine value, and signals that were manufactured to game a ranking system. Once Google could detect the difference at scale, the manufactured version became a liability instead of an asset. That’s the pattern I want you to hold onto, because I think it’s about to repeat.
What’s Happening Right Now in the AI Visibility Space
Over the past year, I’ve watched an entire cottage industry spring up promising to get brands “cited” or “mentioned” by ChatGPT, Gemini, and Perplexity. Some of this is legitimate — genuine digital PR, real earned media, credible expert commentary. But a growing chunk of it is starting to smell exactly like 2011-era link building wearing a new outfit.
I’ve seen pitches for “AI citation packages” that are really just bulk placements on low-authority content farms, the entire pitch being that if a brand name shows up often enough across enough low-quality pages, an AI model will eventually treat it as an established, trustworthy entity. I’ve seen tools that promise to “seed” your brand across forums and Q&A sites specifically because those sites get scraped for training data or pulled into retrieval systems. I’ve even had a client forward me a proposal offering to generate hundreds of fake “expert roundup” mentions, the exact same manipulative tactic Penguin was built to catch, just aimed at a different downstream target.
None of this is a coincidence. It’s the same underlying incentive that drove link spam in the first place: someone figured out a system relies on external signals to judge trustworthiness, and someone else figured out how to manufacture those signals cheaply and at scale. That gap between signal and substance is exactly where manipulation always grows.
Why AI Platforms Are Not Going to Ignore This
Some people I talk to assume AI search is somehow immune to this problem, that because it’s “smarter” than old-school keyword matching, manipulation won’t work or won’t matter. I’d push back on that pretty firmly. If anything, the incentive to solve this problem is even stronger for AI companies than it was for Google circa 2012.
Think about what these platforms are actually selling: trust. Someone asks ChatGPT or Perplexity a question expecting a reliable, unbiased synthesis of the best available information. The entire value proposition collapses the moment users suspect the answers are being gamed by whoever spent the most money seeding fake mentions. That’s an existential risk for these companies in a way it never quite was for a search engine that people already understood was full of ads and SEO gaming.
We’re already seeing early signs of a response. AI platforms increasingly weigh source diversity and cross-verification — checking whether a claim about your brand shows up consistently across genuinely independent, credible sources, rather than trusting any single mention at face value. They favor established, reputable domains for retrieval far more than obscure content farms. And several major platforms have started actively filtering out content that shows patterns consistent with manipulation, including suspiciously uniform phrasing across many low-quality sites, a dead giveaway for bulk-seeded placements.
I’d bet money that within the next year or two, we’ll see something functionally equivalent to a Penguin update aimed specifically at AI citation manipulation. Maybe it won’t have a cute animal name, but the mechanism will be the same: detect manufactured trust signals at scale, and devalue or penalize the sources responsible for them.
The Specific Tactics I’d Steer Well Clear Of
Let me get concrete, because vague warnings don’t help anyone make decisions. A few patterns I’d actively avoid right now if I were running a brand’s AI visibility strategy.
Bulk mention buying on low-quality or clearly irrelevant sites is the most obvious red flag, and honestly the easiest one to spot from the outside too. If a “citation building” vendor can’t explain why a specific placement makes editorial sense for that publication’s actual audience, that’s the tell. Real mentions come from genuine relevance, not inventory availability.
Uniform, templated brand descriptions pushed across dozens of sites are another one I’d avoid. If your company’s boilerplate description appears word-for-word on forty different low-authority domains, that’s not organic reputation building, it’s a pattern any halfway competent detection system will flag immediately, and it’s exactly the kind of over-optimization that got exact-match anchor text penalized a decade ago.
I’d also be cautious about services that promise to manipulate AI training data directly, or that claim insider methods to influence what a specific model “thinks” about your brand. Setting aside whether these claims are even technically real, and I’m skeptical of most of them, the reputational risk of being associated with that kind of tactic if it ever surfaces publicly is not worth whatever short-term visibility bump you might get.
And I’d steer clear of paid guest posts on sites whose only real audience is other SEO agencies and link buyers. This one is a direct callback to Penguin. Those networks were dismantled once, thoroughly, and the domains that still operate this way today tend to carry exactly the kind of low-trust signal profile that both traditional search engines and AI retrieval systems are increasingly good at recognizing.
What Actually Held Up After Penguin, and What Will Hold Up Now
Here’s the part I actually find encouraging, because it’s not all cautionary tale. The brands that came through Penguin relatively unscathed were the ones that had been building genuine authority the whole time, largely by accident of doing things the right way rather than by strategic foresight. They earned links because journalists and bloggers found their research useful. They got cited because their data was original and their expertise was real. That kind of authority doesn’t get penalized when an algorithm gets smarter, it gets rewarded, because it was never dependent on gaming anything in the first place.
I think the exact same principle applies to AI visibility right now, and honestly it’s the advice I keep giving clients even when they’d rather hear a shortcut. Publish content that’s genuinely useful enough that people want to reference it, not because you asked them to, but because it saved them time or answered something nobody else had answered clearly. Build real relationships with journalists, industry publications, and credible sites in your niche, the kind of relationships that produce coverage because your expertise is worth covering. Invest in structured data and technical clarity so that when your content is genuinely worth citing, machines can actually parse and understand it. And be patient, because durable authority, the kind that survives algorithm updates instead of getting wiped out by them, has never been something you can buy in bulk. It has always been earned one credible signal at a time.
There’s also something worth saying about diversification. Brands that put all their eggs in one manipulation basket during the pre-Penguin era got hit hardest, because their entire ranking depended on a scheme rather than a foundation. The same risk applies now. If your entire AI visibility strategy depends on seeded mentions rather than real authority, you’re building on sand, and you should expect the tide to come in eventually.
How We’re Approaching This at Backlinkgen
I won’t pretend this is purely academic for us. We get asked constantly whether we offer “AI citation packages,” and our honest answer is that we don’t sell shortcuts we don’t believe will hold up, because we’ve watched this exact movie before and we know how it ends for the brands that took them.
What we do instead is help clients build the kind of authority that was always going to matter regardless of which platform is doing the ranking. That means genuine digital PR aimed at outlets your actual customers read, not link farms built for search engines. It means content built around original research, real data, and first-hand expertise, because that’s what earns organic citations from both journalists and AI models alike. It means structured data and technical foundations that make your legitimate expertise easy for any system, human or machine, to find and understand. And it means being straight with clients about the fact that there is no reliable shortcut here, because the platforms with the most to lose from manipulation are the ones with the strongest incentive to catch it.
If a vendor is promising to get you cited by AI models fast, cheap, and at scale, I’d ask the same question I used to ask about suspiciously fast link building packages back in 2011: what happens to my brand when this gets caught? Because based on everything I’ve seen so far, it’s not a matter of if, it’s a matter of when.
Final Thought
I don’t say any of this to be alarmist. AI search is a genuine opportunity, and there’s real, sustainable visibility to be won by brands that approach it seriously. But I’ve lived through one cycle of the industry learning this lesson the hard way, and I’d rather help clients skip that particular education this time around. Build real authority, earn real mentions, and let the platforms figure out who’s playing it straight. History suggests they eventually do.
