Skip to main content

How Google and Anthropic Evaluate Content for AI Visibility

Logan Miller · October 28, 2026 · 12 min read

Manipulating search engines with network sites and low-quality automation is no longer effective in modern AI search optimization. Major platforms like Google, Anthropic, and OpenAI are aggressively updating their systems to detect hidden network structures, intent-hiding schemes, and low-effort AI content pipelines. To build sustainable AI visibility across platforms like Claude, ChatGPT, and Perplexity, B2B brands must focus on authentic, highly useful human-driven content and clear regulatory compliance. Data shows that relying heavily on automated text or spamming open APIs yields diminishing returns, especially for high buyer intent queries. Companies that prioritize original insights, direct helpfulness, and source transparency will secure higher trust rankings and consistent citations as LLM citation tracking models continue to evolve.

The End of Hidden Site Networks and Content Manipulation

Logan: Essentially, this is an attempt to manipulate, which is the exact wording that companies, like Google and Anthropic have said that they do not want., you could think about it as your own company too. If you have rules in place and you have a system in place to serve all of your customers and you have one customer that is trying to actively break the rules and manipulate your systems for their benefit regardless of any of the other customers, you wouldn't like that either probably. Anthropic and Google are really just trying to protect their product, which is information. And having people seed incorrect information, one of these methods being that people are starting a bunch of different sites that all lead back to your site, that are not named that they're leading back to your site, that's that's a big issue.

Logan: Now if you essentially say on your different planetary sites that they are a part of your company, but, this is the website for this specific type of content or these specific types of products, etcetera, that's fine. You can interlink between those all you want. But you just. Again, you need about the origins, about the goals, about the authorship, about the ownership. Anthropica and Google really wanna know who is saying what. And so if you're doing that on multiple websites and you're, the secret puppet master behind these multiple websites that are all secretly tied to each other so that they can spin up more links to each other and get more traffic and get more citations, when in all actuality, you're just funneling everyone back to the main website, then they just wanna know about it.

Logan: They just don't wanna play into your little game. And I know that's what a lot of people want, but they will find out sooner or later., you can only run from these types of rules for so long, and Google and Anthropic are some of the most intertwined and high-tech webs that exist on the entire planet, so they will find out.

They're they're not gonna be nice about it when they do either.

Stephen: Let's go into the exact wording of Anthropic and Google's different policies. Their policy prohibits distributing content through networks of fake or ostensibly independent outlets, websites, or accounts to hide that it comes from a common source. Its own example is networks of sites posing as, unaffiliated sources corroborating the same claims. Exactly what we just said, but in different words. Google covers it by saying that, creating multiple sites with the same intent or hidden intent rather of scaling the content and the recommendation or the or the actual, public view of the content of the opinions that you are trying to put out there, and hiding that intent from the public and from Google and Anthropic is really what they don't want.

Stephen: It really is interesting that also Anthropic has said that they ban marketing, selling, or soliciting funding, or any of these, specific products that they have in their list. If you have something that's more of a regulatory product or service, I would definitely recommend going and checking out Anthropix new terms of service because there may be things that you have to say in your blog posts or on your website in order for it to look at your website correctly and be able to cite you, where it should. Next question. Is a web of truth the same as fake source networks like these policies ban?

Logan: Intrinsically, no. By. I use that term web of truth somewhat often when I explain our strategy for optimizing for AI search. The way that I'm talking about a web of truth is creating multiple touch points and multiple pieces of information that do say the same information, but in different ways with different use cases so that they're all useful pieces of content. They're all unique pieces of content. We're talking about our distribution strategy that there's nothing hidden in order to do that.

We're talking about branded content that is on people's YouTube, that is on people's websites, that is on people's social media platforms, that is all being very honest about that it's all coming from the same company.

Logan: We want to make sure that they're known as the source very specifically, which is the opposite of what these fake networks of sites are doing. They're purposely trying to hide the identity of the authors and hide whose perspective it's coming from so that it just looks like there's a bunch of different people talking about it. What we're doing instead is just trying to help the customer more by saying, if you want this information in long form video, we're going to give it to you there. If you want this information in short bite sized videos that really just answer very specific questions for you, we're gonna put it there. If you want it on social media, like Instagram or threads or x or Facebook or TikTok, we're gonna put it there.

Logan: If you want it in written formats, you can more easily just peruse through the information and have a deeper dive. We're gonna put it there. But we're always going to brand it with our companies or our clients' names so that everyone knows who's the author. And for us, that's a benefit because we want people to know who the author is so that they can know to credit the clients with that information. And when you have more sources and you show yourself as more helpful to the consumer and being willing to serve your customers no matter where they want to be served, then you're going to have a higher trust ranking within these systems, within Claude and Google, but also within ChatGPT and Grok and, and Perplexity and Copilot and all the others.

Watch the full interview on YouTube

Updating Compliance for Anthropic and Modern LLM Terms

Logan: That's what our strategy is, and that's what we talk about as our web of truth, which is really important for us to differentiate that from a fake network of sources or fake network of sites that these specific, new terms of service from Anthropic are banning. We're almost intrinsically the opposite of that by making sure that Anthropic and Google and everyone else know exactly who the authors of the content is.

Stephen: Big question that comes up a lot. How do I show up in AI search without getting flagged as spam?

Logan: Well, the easiest way to not be flagged as spam is to not spam. Another easy way to not get flagged as spam is just not to use AI generated content. Now a lot of people are gonna get mad at me for saying that because we have very strong views on AI generated content here at Attractify.

Watch the short: How to Show Up in AI Search Without Getting Flagged

What Kind of Expert Content Gets Cited by AI Search?

Logan: We don't use it at all. We never have. Because as the Internet gets more and more saturated with AI content and AI generated text, we have seen over time that what Google and Anthropic and ChatGPT and all the other LMs want is more useful human generated content that they can't get from anywhere else. And they want it in a way that's easy to digest, and they want it in a way that is not somewhere else on the Internet. They've said that time and time again, and they've also proved it with. Or proven it in their actions. Every single time that Google does a spam update, every single time that Anthropic or ChatGPT or OpenAI or any of these other update their terms of service, they get more and more strict on AI generated content.

Logan: While you might be able to have 30 to 40% of your content generated by AI as long as you're not publishing past a certain amount of words or pages per week or month or anything right now, that's gonna become less and less valuable over time.

Watch this moment on YouTube (26:15)

The more and more strict they get and the more content that gets generated by AI and put into the Internet that they'll know about, they'll be able to see if it's AI or not, the more valuable that human generated content becomes because it becomes a much smaller and smaller portion of all the information that they have access to. While our customers are currently getting their results with human generated content, they're also going to keep getting those results because as the rest of the Internet becomes more AI, theirs is going to stand out as less AI and is going to be a much rarer commodity percentage wise of the content.

Why Scaled AI Content Yields Diminishing Returns

Logan: And, also, our content is very directly answering the exact questions and is being optimized to be able to be read and digested as quickly and easily as possible by LLMs. That's how we win. It's it's. I like to say the four c's. It's content, context, constant, and coverage. We create the correct content that answers the actual questions. We give it the right context by structuring it in a technical way and placing it with other content, linking it to other content that is pertinent so that AI has an easier time understanding it. We keep it constant by making this flow of information and content be pumped out in a very predictable manner that, again, search engines and LLMs can just always be refreshing that content.

Logan: We're updating pages even without making duplicates. And then we're putting it out. We're distributing it. That's the coverage part. We're distributing it in all of the different places.

People used to say that Reddit was a really great place to get citations from, and it was. However, we saw the writing on the wall that was not going to be for forever because the API that Reddit put out there was way too easy to spam with bots, and we just thought that it was gonna get super diluted and useless pretty quickly. And now on August 13, AI. Or excuse me, OpenAI, ChatGPT, publicly announced that Reddit was no longer being heavily. As heavily cited at all. We found it in our own data as well.

Logan: We saw a drop off of Reddit citations specifically for high buyer intent queries far before that. They were never, in my opinion, and our data shows it, they were never the most heavily cited source for high buyer intent queries, at least in the industries that we were running citations and thousands of citation tests per week on.

Watch this moment on YouTube (28:56)

As the as the appetite for different types of content shifts, we are staying relevant within these LLMs because we're putting out written content. We're putting out off-site content. We're putting out videos. We're putting out long videos. We're putting out short videos. We're putting very specific, pullouts of the quotes of these transcripts, as the actual captions of these videos and the descriptions and the titles.

Logan: We're we're placing the content everywhere. And then no matter what happens, whether it starts liking video more or it starts liking written more or it starts liking actual on-site content more or off-site content more, our growth line stays pretty consistent. We don't experience the same dips that a lot of these other AI generated platforms and AEO and GEO agencies and softwares do, and it's because of that.

Comparison at a glance

Fake source network vs Attractify's "web of truth" distribution strategy
FactorFake Source NetworkWeb of Truth
AuthorshipDeliberately hidden so it looks like many independent peopleEvery asset branded with the client's name
Purpose of repetitionCorroborate the same claim from ostensibly unaffiliated outletsSame information in different formats for different use cases
Formats usedSites, accounts, sock puppet Reddit posts, fake reviewsLong and short video, YouTube, Instagram, Threads, X, Facebook, TikTok, written content
Policy statusExplicitly prohibited by Anthropic's new termsCompliant; makes the client the verifiable source
Effect on trust rankingRisk of throttling, suspension, or terminationHigher trust ranking across Claude, Google, ChatGPT, Grok, Perplexity, Copilot

Frequently Asked Questions

How do Google and Anthropic handle hidden site networks?

Both Google and Anthropic actively penalize entities that create multiple websites with hidden intents to manipulate traffic or artificially inflate links and citations. These platforms use high-tech detection systems to identify when multiple domains are secretly controlled by a single puppet master. Instead of attempting to cheat the algorithm, brands build sustainable trust by maintaining clear intent and serving accurate content directly to their audiences.

How do I get cited by ChatGPT and Anthropic's Claude?

To earn citations in top LLMs, focus on producing rare, highly authoritative content that cannot be easily generated by AI. Demonstrating broad utility, adhering to platform terms of service, and clear regulatory compliance builds your brand's underlying trust score across systems like Claude, ChatGPT, Grok, and Perplexity.

Does Reddit content affect AI search rankings for high buyer intent queries?

While Reddit data has been heavily used by AI models, proprietary data shows it has rarely been the most cited source for high buyer intent B2B queries. Because public APIs and forum networks are frequently diluted with bot traffic and automated spam, LLMs increasingly rely on verified, original authoritative sites for revenue-critical recommendations.

What percentage of blog content can be generated by AI without losing trust?

While publishing 30% to 40% AI-generated content may pass current volume filters, relying on synthetic text will offer diminishing returns over time. As the web becomes oversaturated with AI text, human-generated content becomes a rarer and more valuable commodity. Search engines and AI engines prioritize original human insights over recycled model outputs.

Why is Anthropic changing its terms of service for content recommendations?

Anthropic updates its terms of service to refine how Claude indexes, evaluates, and cites regulatory and commercial content. Brands in highly regulated or technical spaces must stay aligned with these structural guidelines to ensure their content is parsed correctly and cited accurately when users ask for B2B product and service recommendations.

Get your free AI visibility scan

Find out where you stand in ChatGPT, Perplexity, Gemini and Google's AI answers.

Run My Free Scan

← Back to the blog