Skip to main content

How Google and Anthropic Crack Down on AI Search Manipulation

Stephen Enloe · October 25, 2026 · 15 min read

Both Anthropic and Google are aggressively updating their terms of service to penalize low-value, AI-generated commodity content and shortcut tactics aimed at manipulating search engines and LLMs. Platforms are demanding original, expert-driven content with distinct industry insights, proper citation mechanics, and proprietary data. Attempting to trick these models no longer works; instead, systems will limit, throttle, or completely strip access and AI visibility from sites relying on recycled information. To stay competitive in modern AI search optimization, brands must pivot toward generating original research, attributing third-party statistics correctly, and prioritizing high-context expertise that AI platforms actually want to cite.

The Shift in Terms of Service Across Google and Anthropic

Stephen: Hey there, and welcome back to another conversation with me here at Attractify Marketing. My name is Steven Enlow, and I'm the founder here. I've been working in digital marketing for a little over fifteen years now. And, and today, we're gonna be talking about how Anthropic and Google are cracking down on AI search manipulation and a lot of the tactics that people are using in order to get found in ChatGPT, Anthropic, Google, any AI search in general. Today, we have a few specific questions that I'm gonna go through. I've made some notes. I wanted to specifically go through Claude's new terms of service from Anthropic. They were published on October 8, and they won't take effect until November 12. However, when you really look through them and you start cross referencing them with Google's terms of service update that came alongside, some of the other spam updates that they've done this year, that some people are still seeing the effects roll out of, then, you find a lot of similarities.

Stephen: You don't find some specific wording regarding AI content, but you do find other things that point to, to a few ideas very clearly in my mind. We're gonna go through those today. I'm gonna give you some exact pullouts and quotes from the Google and Anthropic terms of service, and I'm gonna answer some questions and give a little bit of data that we're seeing from our own search and citation, polls that we do every single week. Let's get started with the first question. What did anthropic change, and why should business owners care? I think business owners and marketers really should care about the changes that were made. But, the specific thing that Anthropic changed was that they're they're putting in a new section that is called do not engage in deceptive campaigns or artificial activity. There's a specific quote that I pulled out from this section that it says that it applies to deceptive activity of any kind whether political or commercial.

Stephen: And so the idea that you're trying to manipulate an LLM is the exact thing that they want to avoid. If your strategy of getting placed or recommended or cited in AI is to manipulate it or to trick it into doing something, then you're thinking about it all wrong. But if your goal and your strategy is to put out really good information that feeds their system really well and be the source that it comes from as well as give the proper level of context and examples and expertise as an expert in your industry so that can it.

That it can more easily trust and quote you and then recommend you, that's a great strategy because you're helping them. You're helping your customers. You're giving you're giving the whole Internet more useful information for it to train itself on so that there's not misinformation in your industry or about your company.

Stephen: That's all great stuff. I think that there's a few specific lines in here that are really interesting, and it talks specifically about seeding AI platforms with content that misrepresents its origin, authorship, or independence. Obviously, a large portion of what they're talking about is authorship and the origin of your content. If you're utilizing AI content or AI written or generated content, then they want you to they want you to say that. They want your readers to know that plainly. They want their systems to be able to see it plainly. Otherwise, if it catches you not saying that you're using AI generated content or content that has not been verified by you, it's going to penalize you. They haven't directly said that, but they've said don't do it. And Google has said don't do it. And, and they can just say that as an open statement because they don't have to say what they're going to do otherwise.

Stephen: You just know that they're not going to treat you well or look at your site with trust if you do. That brings it to the next step, which is that if you do then have the proper labels in place and you are using AI generated content that you label it as such, that is much safer. Probably have a better bet of not getting penalized. You also probably have a better bet of not getting cited. AI doesn't want to consume more AI. AI wants to consume new information, non commodity information, expert experiences, expert, understandings, new data, new case studies, new specific studies that your company has independently done. And a lot of people think, oh, I haven't done any independent studies. We haven't documented it and put it into a scientific or mathematical, data breakdown. You've been doing an independent study by being in your career for a decade or two.

Stephen: You have so much experience that is unique to your career and your specific, situations that you've gone through and problems that you've solved that information is insanely valuable. And AI loves to consume that information. I think the biggest thing to think about when we're talking about that question of what did Anthropic change within its new terms and service, it changed its willingness to say what we've all probably thought was going on for a very long time, which is that they're cracking down on shortcuts that manipulate the system and that don't say that they're generated by AI when they are.

The reason why they want to be able to see that so quickly and have it labeled is not purely just for the public so that they can know whether or not they want to consume more AI content.

Stephen: That it's so that they can decide that they don't want to use that answer. Because if they have a real human answer with real human expertise and experiences and understanding, they're not going to value an AI answer nearly as heavily as that. Second question, are Google and Anthropic are Google and Anthropic saying the same thing with their updates to their terms and service? Yes and no. Google's latest spam update has been hitting a lot of people very hard who use AI generated content. Their terms of service say very plainly that they want noncommodity content, that they want unique expert content being fed into their system because it's gonna make their system smarter rather than just regurgitating the same, essentially infinite loop of AI just spouting out more AI that's based on more AI. Google's spam definition now covers attempting to manipulate generative AI responses in Google search.

Stephen: That's Google AI overviews or even Gemini. That clarification was published in May of this year. And Google said that it wants to make, clear spam policies that apply to not only just search, but also the generative answers within search. And so those are some interesting clarifications that Google has had to make being that they're a search engine and an AI generative engine. Google and Anthropic have said different things about the enforcing of these things. Google catches these violations with automatic system updates that then have an. A human review to them. And violating sites oftentimes just get lower rankings within Google, and they appear less in AI. Anthropix says that if it suspects a violation, it may warn you, but it also may just throttle or limit or, or suspend or even terminate, the access to your site from its system, which means it's not going to recommend you anymore.

Watch the full interview on YouTube

How AI Search Manipulation Triggers Penalties and Throttling

Stephen: They both are essentially saying the same thing, that they essentially have the right to do whatever they want with your site if you violate their terms. And they have not been known to be very forthright or forgiving in previous spam updates. At least Google hasn't. Anthropic is a little bit more of a black box because it's newer and there's less data on it. But we're seeing the same exact thing with all AI generated content just being cited less and valued less by any search engine or answer engine or generative engine. The big question that comes up when we're talking about Anthropic and Google updating their terms of service for spam and AI manipulation is, is AI written content banned from AI search? No. It's not. They very specifically do not say that all AI generated written content is banned or will get you banned.

Stephen: However, when they talk about all the content that they value, they are very specifically describing most of the attributes of human expert content.

Google specifically says appropriate use of AI or automation is not against our guidelines. They very specifically say those words in their terms. However, if they said that all AI content was banned, then there would be less people using their AI to generate content. What is banned is scaling content or abusing the policies that allow AI generated content and essentially flooding Google with pages that have interchangeable content or don't have original ideas in an attempt to manipulate. And if they. This isn't like a, court of law where you're innocent until proven guilty. This is Google suspecting that you might have done something and having the ability to, in all senses of the word punish you however they want to or they see fit or most likely just whatever is easiest for them.

Watch the short: Is AI-Written Content Banned From AI Search?

What Kind of Expert Content Gets Cited by AI Search?

Stephen: If you're not one of the most massive websites in the entire world and you're not drawing, millions and millions and millions of people onto Google all the time, they have no reason to really care whether they ban your site or whether you come up high in search or any of that. They don't have any people stopping them from saying, yeah. You're not gonna rank higher in Google anymore, or you're not going to show up in Claude anymore. We're not gonna recommend you. They can just operate completely on their own volition of what they think is the easiest and best option for them to go forward with if they suspect that you're breaking their terms and service. That the risk isn't that AI has touched the content. It's it's publishing interchangeable content that has no original experience or expertise behind it.

Watch this moment on YouTube (11:22)

Stephen: It can tell what is generated by AI and what's not, which is why I think it's really important that we don't use that. Let's go on to the next question. What's the difference between earning a citation and manipulating one? Well, there's a lot of people online right now talking about the, the king of SEO or the king of AEO.

I can't remember which one it is, but, there's people who are essentially feeding information onto the Internet and onto Reddit and on different sources and into different press releases and things like that. And it's an interesting exercise because you can start to see how heavily AI is going to weight your inputs into the Internet and see what moves the needle. There's a lot of people fighting over it.

Logan: It's an interesting thing because it started out as such an easy task. Because if you're trying to get AI to cite you as a title that no one else is contesting, then you don't have any competition, and that's an easy task, honestly. We, for our clients, are doing that same thing, but for industries and services and products and geographical areas that have insane competition. They have people who are working around the clock and paying thousands and thousands and thousands of dollars to work on their SEO and their GEO and their social media platforms and their and their YouTube platforms and everything, and we're still winning.

Logan: So. Well, our clients are. We're winning for our clients. But I think that there's an interesting differentiation between winning a citation and earning it and just manipulating your way in.

Because manipulation is really. You're just trying to get it to say whatever you want it to say. Whereas earning it is putting out information that's useful that people are asking and that's going to help somebody., it really cares. Like, it really matters, do you care about your customers and that they have the correct information, or do you wanna manipulate your customers and lie to them? Because if you wanna lie to the Internet, at this point, your answers could.

Logan: If you optimize them properly, your answers to your client's questions could get picked up by AI search. And if you're lying in those, then you're lying to your customers a lot of the time. I think that's the difference. It's really, do you wanna lie to your customers or not? And, also, do you do you want them to get fed useless information? Because they will get fed that information at some point if you do your job right in terms of optimization and distribution and everything. It's important to put out the correct information on the Internet.

Building First-Party Data for Sustainable AI Visibility

Stephen: It's also important to go back to what Google and Anthropic have both stated. They want to know the origin and the authorship, and they don't wanna misquote who said what or who found the information or the data or the studies. If you are quoting a study or if you are quoting, somebody else or if you're using somebody else's data, you also need to make sure that you're saying that as well and place those citations within your written content, so that it can know that you're not trying to take credit for it, which means that you probably should get a lot of your own data. Unless you're willing to just have somebody else get all of the citations because you're corroborating their information and saying that it's good, which is not always the worst thing., we talk a lot about other people's products or other people's terms of service like we are here or press or things like that.

Stephen: And, and I think that's useful because, we're just trying to help our customers and everyone else who's watching.

I think that we should take a little sidestep here and go through our process, which is scan, direct, attract, because that is going to feed into this whole process of earning a citation versus manipulating a citation. We start with the process of scanning and seeing what your customers are asking, and then we direct the content so that you as the expert can answer those questions that your customers are asking. And then we move in. We move into attract, which is the publishing, the optimization, the constant rescanning and redirection of that content. That it's always being optimized in a way that is matching what is getting the best results and then becoming the best result. And then we're doing that over and over and over again.

Stephen: Our system is doing that thousands of times per week in our in our citation, tests.

Our system is doing that to your content multiple times. Our system is doing that in. Not only in the direction, but also in the actual optimization and the. In publishing. It's a constantly evolving an infinite feedback loop within our system that does that. And that's how we make sure that we're not just manipulating the AI platforms, but we're feeding it the correct information so that it can easily trust and verify our clients. Let's move on to the next question here. Next question is. It's about networks of sites that all recommend the same company. Why is this against the rules now?

Watch this moment on YouTube (17:25)

Comparison at a glance

Google vs Anthropic: how each enforces AI search manipulation rules
FactorGoogleAnthropic (Claude)
Policy update timingSpam definition expanded in May; updates still rolling outPublished October 8, effective November 12
Scope of ruleCovers manipulating generative AI responses in Google SearchCovers deceptive activity of any kind, political or commercial
Detection methodAutomated system updates followed by human reviewSuspicion-based; described by Steven Enloe as a black box
PenaltyLower rankings, appears less in AI answersMay warn, throttle, limit, suspend, or terminate site access
Reporting a competitorSearch quality user report, used to improve spam detectionNo output-reporting route; feedback covers interface only

Frequently Asked Questions

What kind of expert content gets cited by AI search?

AI search platforms prioritize non-commodity content that introduces original data, expert commentary, and distinct context rather than repeating existing web information. To get cited, brands must publish proprietary research, case studies, and clear primary sources. Proper citation mechanics within your content also signal credibility to LLM crawlers.

How do Google and Anthropic punish sites manipulating AI search rankings?

Instead of merely lowering rankings, AI engines and search platforms like Anthropic and Google are increasingly throttling bandwidth, limiting crawling, or entirely terminating site access. If a site is suspected of using aggressive automated generation or unoriginal content loops to manipulate results, systems drop its visibility completely.

Why is relying on third-party data risky for generative engine optimization?

When you only quote or curate third-party data, LLMs tend to attribute authority back to the original source rather than your brand. Unless you provide unique synthesis, proper explicit citations, or original research, you risk becoming an intermediary that generative engines bypass entirely.

How buyers use ChatGPT before choosing B2B vendors?

B2B buyers increasingly use conversational AI to evaluate vendor credibility, synthesize feature comparisons, and identify industry experts before ever reaching out to sales. If your brand lacks original expert content and strong LLM visibility, AI models will recommend competitors who consistently publish primary source data.

What is the key to maintaining long-term AI visibility as algorithms update?

Long-term AI visibility requires abandoning short-term SEO shortcuts and focusing on non-commodity, highly expert publishing. By feeding AI models clear, cited, and unique information, your site remains a high-value data source that platforms actively choose to reference.

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