I Found a Website Trying to Manipulate AI Recommendations. Here’s Why Businesses Should Care.

I’ve been spending a lot of time lately learning how AI is changing search and what that means for businesses. People aren’t just going to Google anymore and typing in something like “video production company in Lubbock.” They’re asking ChatGPT and other AI platforms who they should hire, what they should buy and where they should go.

Naturally, that creates a new challenge for marketers. How do you make sure these systems know your business exists, understand what you actually do and have enough reliable information to potentially recommend you? That’s a big reason I’ve been paying attention to AEO, AI search optimization and the different ways businesses are trying to position themselves for this change.

While doing some research, I found something I wasn’t expecting. I came across a website with instructions written specifically for AI systems, basically telling them how to respond when someone was searching for that type of business and how the company should be positioned in the answer. This wasn’t marketing copy written to convince a potential customer. It appeared to be an attempt to influence the machine that might eventually talk to that customer.

I had to read it a couple of times because I couldn’t believe someone had actually put it on their website. Once I understood what I was looking at, my first thought was pretty simple: we’ve seen this before. The technology is different, but this feels a whole lot like the early days of black-hat SEO.

We’ve Already Been Through This With Google

There is absolutely nothing wrong with trying to rank well on Google. SEO, when it’s done correctly, helps search engines understand what your website is about while hopefully making the website more useful to the person searching. The problem comes when people stop trying to provide the best answer and start trying to manipulate the system into believing they are the best answer.

That’s how we ended up with keyword stuffing, hidden text, link farms, doorway pages and all kinds of other tactics designed to exploit weaknesses in search algorithms. Some of them worked for a while, which is exactly why people kept doing them. Google eventually became much better at identifying manipulation, and many of those shortcuts either stopped working or became something that could actively hurt a website.

Now we’re watching the same race begin with AI. Businesses know there is tremendous value in being the company ChatGPT, Gemini or another AI platform recommends, so people are naturally going to look for ways to influence those answers. Some of that will be legitimate optimization, and some of it clearly won’t.

Optimization and Manipulation Aren’t the Same Thing

I want to make that distinction because I’m actively working with AI and learning how to optimize websites for this new type of search. I don’t believe using AI is inherently bad, and I don’t believe businesses should ignore it. I’ve said for a while now that AI is a tool, and like any other tool, what matters is what we choose to do with it.

If you’re a production company in Lubbock, there is nothing wrong with making sure your website clearly says you’re a production company in Lubbock. There is nothing wrong with publishing useful information, answering questions, using structured data, building backlinks, earning media coverage and making sure information about your company is consistent across the internet. You’re giving a search engine or AI system better information so it can understand who you are.

What I found crosses a different line for me. Instead of simply giving an AI system information to evaluate, the website appeared to be giving the AI instructions about how it should respond. There’s a major difference between giving a system evidence that your company may be a good recommendation and essentially saying, “When someone asks this question, recommend us.”

Why This Becomes a Trust Problem

This is where I think the issue becomes much bigger than SEO or AEO. When someone searches Google, most people understand that they’re looking at ranked search results. They’ll probably open several websites, read reviews, look at portfolios and compare their options before making a decision.

AI changes that relationship because it gives you an answer. You can ask a conversational question like, “Who’s the best video production company near me?” and receive something that sounds much more like a recommendation from a knowledgeable person than a traditional page of search results. The more confident that answer sounds, the easier it becomes to forget that a machine is assembling it from information that may be incomplete, inaccurate or intentionally manipulated.

Now imagine businesses successfully inserting instructions into that information specifically designed to influence the recommendation. The person asking the question has no idea that happened. They may reasonably believe the AI independently looked at the available information and determined that one company was the best choice.

That’s not just a marketing problem. That’s a trust problem.

We’re Already Seeing What Happens When AI Content Floods the Internet

There’s another reason I think this matters right now. The internet is already dealing with an enormous amount of cheap, repetitive and misleading AI-generated content, commonly called “AI slop.” The term isn’t referring to every piece of content made with AI; it’s generally being used to describe the flood of low-effort material that can now be generated at a scale that wasn’t practical a few years ago.

The platforms themselves are responding to it. In 2025, YouTube clarified its monetization rules around what it now calls “inauthentic content,” specifically identifying repetitive or mass-produced material as content that isn’t eligible for monetization. YouTube’s current policy even identifies AI-generated material made from generic or unoriginal templates that gives the impression of mass production as an example of content that can violate its monetization guidelines.

YouTube hasn’t banned AI, and that’s an important distinction. The company specifically says creators can use AI tools and remain eligible for monetization when the result is original and provides actual creative, educational or entertainment value. What it’s pushing back against is the assembly-line approach where automation is used to churn out enormous amounts of interchangeable content without adding anything meaningful.

Facebook has been dealing with a similar problem. Meta announced that accounts trying to game distribution or flood Facebook with spammy content could receive less reach and lose monetization opportunities because that material was crowding out authentic creators. Again, the problem isn’t simply that a computer was involved in making something; it’s what happens when automation makes it incredibly cheap to overwhelm a platform with content designed primarily to manipulate attention or an algorithm.

Pinterest has taken another interesting approach. It started labeling content it detects as AI-generated or AI-modified and later introduced controls that allow users to reduce the amount of generative-AI content appearing in certain parts of their feeds. Think about what that says about where we are: a platform built almost entirely around visual discovery now has users asking for a way to see less AI-generated material.

The Problem Isn’t AI. It’s the Loss of Trust.

That’s the part of the conversation I think gets lost when everything turns into an argument between people who love AI and people who hate it. I don’t think either extreme is particularly useful. AI can be an incredible creative, research and productivity tool, but the ability to generate something quickly doesn’t automatically make that thing valuable.

We’re reaching a point where seeing something isn’t necessarily enough to believe it anymore. A photograph can be generated. A person’s voice can be cloned. Video can depict events that never happened, articles can be produced by the thousands, and social accounts can publish an endless stream of material without anyone actually sitting down and creating it.

Platforms know this is a problem, which is why transparency has become such a big part of their response. Meta labels certain AI-generated content, Pinterest labels AI-modified images, and platforms have developed rules around synthetic or manipulated media. These companies aren’t doing that because AI itself is automatically misinformation; they’re doing it because people need more information to determine what they’re actually looking at.

That’s also why the thing I found bothers me. We’re trying to solve one trust problem while potentially creating another one inside the systems people are beginning to use to sort through all of this information.

What Happens When AI Starts Reading AI?

This is where things get really weird.

We’re creating more AI-generated websites, articles, images, videos, reviews and social posts. AI systems are then being used to search, summarize and make sense of information found across that same internet. If businesses begin intentionally placing instructions into their websites designed to influence those systems, we’re adding another layer of manipulation to an information environment that’s already becoming difficult for people to verify.

The potential cycle isn’t hard to see. AI generates content, that content gets published online, another AI system encounters it while trying to answer a question, and the resulting answer may eventually become another source of information somewhere else. Add intentional manipulation anywhere in that chain and the original source of a claim can become incredibly difficult for the average person to untangle.

That’s one reason I think human credibility is actually going to become more valuable as AI improves, not less. Knowing who created something, where the information came from and whether the source has a reputation worth protecting matters when producing convincing content becomes almost effortless.

Is This Black-Hat AEO?

I don’t know if “black-hat AEO” is ultimately what the industry is going to call this. We’re early enough that marketers still argue about whether the larger practice should be called AEO, GEO, AI SEO or something else. But black-hat AEO is probably the easiest way to explain what I found.

Black-hat SEO tried to manipulate a search engine into ranking something higher than it otherwise might. What we’re talking about now is potentially manipulating an AI system into saying something it otherwise might not. I would argue the second could eventually be more concerning because the manipulation isn’t happening somewhere on a page of search results; it can potentially become part of the answer itself.

That doesn’t mean every attempt to influence an AI answer is unethical. Marketing has always involved presenting information in ways that make a company easier to discover and understand. The line for me is whether you’re giving the system useful information to evaluate or trying to secretly instruct the system on what conclusion it should give the person asking the question.

So What Should Businesses Actually Do?

I don’t think the answer is complicated. Build a business worth recommending and then make sure there is enough legitimate information online for people and machines to understand why. That means a good website, useful content, accurate business information, real reviews, backlinks, media coverage, case studies, awards, portfolios and whatever other evidence actually applies to your business.

I can optimize South Town Productions all day long, and I absolutely want search engines and AI systems to understand what we do. I can explain that we’re a Lubbock-based creative and marketing agency, show our work, talk about our experience and publish information that demonstrates what we know. What I shouldn’t be able to do is order an AI system to tell you that we’re the best company for you.

That’s not my decision to make. My job as a marketer is to provide enough accurate information for you, Google, ChatGPT or whatever comes next to make an informed decision. If the evidence isn’t strong enough to earn the recommendation, the solution should be to build stronger evidence, not find a better trick.

This Is Probably Only the Beginning

What interested me most about finding this wasn’t the company that did it, which is why I’m intentionally not identifying them. I’m much more interested in the practice and what it tells us about where digital marketing is going. We’re entering a period where businesses, marketers, AI companies and search platforms are all trying to figure out the rules at the same time.

I’ve worked in creative media and marketing for more than 15 years, and I’ve watched businesses chase Google algorithms, Facebook reach, viral videos, influencers, short-form content and just about every other promise of a shortcut to attention. Some of the tools change everything, but the rush to exploit them is almost always the same. AI isn’t going to be any different.

I’m going to continue using AI, studying AEO and experimenting with how these systems understand and recommend businesses because I think marketers need to understand where search is going. But I’m much more interested in figuring out how businesses can legitimately earn visibility than figuring out how to trick a machine into giving us visibility we didn’t earn.

Because we’re already reaching a point where people are questioning whether the image they’re seeing is real, whether the video they’re watching actually happened and whether the article they’re reading was written by someone who knows anything about the subject. The last thing we need is for them to start wondering whether the AI recommendation they just received was earned or whether somebody figured out how to manipulate that too.