Google’s New AI Spam Detector Could Wipe Out Entire SEO Networks

If you're using common tools to automate a ton of SEO writing, this is something worth considering. And it's also interesting for accounts using mass amounts of AI video to do SEO. And this has become a common spam tactic. This is how Google is thinking about catching AI writing used for SEO spam. Glenn Gabe shared this on X. He said, "Like I've said before, there is no way Google lets companies spamming for AI search continue. They will ramp up systems plus add manual actions. Google research shows how AI spam can be detected." So, he shared this article on Search Engine Journal by Roger Montti, who is awesome. The article is titled Google research shows how AI spam can be detected, but it also has the subtitle, the nuance, "Google research shows why AI-generated spam is becoming harder to catch and why content-level quality filters may no longer be enough." And so, it talks about what Google is using past content-level quality filters. And I read through this, I cherry-picked out the best lines, and that's what I got for you on this episode of the show. So, if you want to see how Google is catching AI SEO spam, this is it. Google researchers published a new paper detailing a new way to catch spammers who are using generative AI to flood Google's platform with spam and overwhelm its quality filters. While the research is focused on identifying video content spam, the techniques described could give an idea of the methods that Google could use for web content spam. In fact, the research paper discusses a text-based generative AI identification system. The new system is said to be a highly accurate defense against coordinated generative AI spam, which means that something like this could conceivably be in use. The new system is called scalable cluster termination system, and the research paper, which is called scalable detection of adversarial synthetic slop and coordinated media abuse, a Laura-enabled multimodal defense system. They're using the term slop in the in the research paper title. The system succeeds because it looks for the organizational structure of an attack, which is the mass reuse of a specific semantic narrative template instead of evaluating isolated videos one by one. So, you can imagine a similar thing with text across a large amount of pages or sites, looking for, again, the mass reuse of a specific semantic narrative template. The research paper also describes the use of text embeddings, salient terms, and templated narratives as a part of their content classifier. If a high percentage of accounts in an infrastructure cluster are identified as using the same AI-generated text/media templates, the entire cluster is terminated. The paper says that when attackers adopt new generative models, Google can adapt its synthetic spam detection system faster by using low-rank adaptation, LoRA, and automatic prompt optimization, APO, instead of retraining a massive AI model. And this is a quote from Google now, "Our approach allows for the efficient adaptation of the large proprietary LLM without the prohibitive computational cost of full fine-tuning. It allows us to engineer prompts that adapt to new, quote unquote, slop trends faster than retraining a dense model." So, this is Google becoming a lot more agile at detecting and preventing AI spam. The researchers acknowledge the use of Sentence BERT, SBERT, as a way to identify semantically similar sentences. They cite Sentence BERT to validate a core assumption of their paper. The core assumption is that automated AI-generated text leaves a distinct mathematical footprint, text embeddings that can be detected. So, the core assumption of their paper says, "Automated AI generated text leaves a distinct mathematical footprint that can be detected." It could be that sentence BERT has only recently been used by search engines like Google for catching AI generated text spam, only recently. Now, this is where the article gets into how generative AI spam has gotten out of control and could overwhelm the current methods, which is why these new methods are necessary. The researchers identify three reasons why generative AI spam is out of control and overwhelming current methods for detecting low-quality content. Here's another quote from Google, "Online video platforms face an exponential challenge in detecting and mitigating the flood of AI generated slop and synthetic spam perpetuated by coordinated malicious actors. This content is increasingly designed to exploit the limitations of traditional media forensics, often utilizing generative AI to produce unique localized variations of harmful or low-quality material at scale. Traditional content-centric moderation fails against this coordinated adversarial generation strategy." Again, that's from Google researchers. The article continues, they're talking about spammers deploying infinitely unique content that is functionally identical as a way of getting around traditional content analysis and mitigation strategies. This is precisely why they're zooming out to look at clusters of accounts to identify the actual fingerprints of the spammers or their automation. The research paper is focused on identifying AI generated video spam, but it begs the question, can something like this be used to identify AI generated text spam? Instead of looking at a single suspicious video in isolation, the system uses a two-pronged machine learning approach to spot entire networks of automated accounts, botnets, that are flooding the platform with low-quality AI-generated spam. Their test data shows that the system results in significant impact in catching clusters of spam with a high level of accuracy. And remember, Google has YouTube and it has everything that it has crawled and penalized and indexed and removed from the index in Google search, which is I mean, a ridiculous amount. All the data to find patterns from. Lily Ray shared Glenn Gabe's take on this and said, "Wonder if AI content tools are concerned at all about this." Which really makes sense because you could see similar patterns of writing and publishing velocity among many sites that are using AI content tools. And then I want to go into the comments of Glenn Gabe's post because there's a common question that I see a lot of people having, asked by Orit Mutsnik, who said, "Didn't Google say they don't have a problem with AI-generated content as long as it provides value? And if all AI-generated content is considered slop, why did they wipe out all those small publishers sharing their own unique experiences?" So, Glenn said, "This is referring to AI spam and slop, not all AI-generated content. But scaling thin or low-quality content, including doing that via AI, is super dangerous. I have covered that a million times. I believe you have seen a few of those posts and shares." Orit said, "Yes, but these systems identify all AI-generated content and to this they apply their definition of spam. So, you could still produce AI-generated content with a human in the loop and find that it has value to your standards, but Google might disagree because it's AI and not meeting their standards." Glenn said, "They use multiple signals, not just one. Just having AI-generated content would not mean spam or slop. There's never one smoking gun, but again, I would caution any site heavily using AI-generated content to take a serious look at what they are doing, evaluate if it's truly high quality, if it provides value, if it's insightful, etc. If that turns into scaled content abuse, look out. So, keep in mind, you could be using AI to make content, and it's good content, and it's satisfying search intent really fast, and it has great user signals, and you're going to be fine, maybe, and you're probably not publishing a billion times a month, though you could if it's good enough, in theory, if it's good enough, which it likely would would not be. But, you know, it's also skirting a fine line. It's something though that is really worth keeping in mind for companies who are considering using these AI content tools or just going really, really hard with AI SEO content on a money site. As for me, I'm not going to do this because I like playing the long game with a single brand, building up a single brand on a strong domain. But, but I got pitched a guy who has 500,000 AI SEO spam sites, and I think he wants to come on this podcast, and that's going to be a super interesting conversation. So, I'm going to invite him on the show. 500,000 AI SEO spam sites. You know, I had someone with 9,000 sites on the show. That was a really fun conversation, and then I had on somebody with 200,000 domains and 30,000 sites on the show. That was also a really fun conversation. And now, I'm going to have somebody with websites on the podcast, and I imagine that is going to be an insane conversation. Anyway, this was a fun share. It was a fun read. Roger Monti always writes great stuff. Glenn Gabe and Lily Ray also always share great stuff. If you want to learn my specific method of SEO, it's that works I just blown my mind how well it works. I shared on X my SEO landing pages with my method of SEO are converting visitors from search to my product by 68.3% so for example I have an app that has freemium you can just use it free with no sign up required if you want more features then you have to start paying we have SEO landing pages targeting people who are literally looking for what our app does and people coming to these pages from search seven out of 10 of them are clicking through to our app that's insane that is an insane conversion and my method of SEO which I show in combat keywords my SEO course is getting great testimonials from other people it works in all niches local e-com I was looking at an e-com site from somebody who got the course who's doing really well and I I just couldn't believe what I was seeing this person just has this crazy spike up in organic traffic ranking for really valuable purchase intent terms I am inspired by what people are doing with the course and so if you want that for your business I would really recommend it I spent a year making it constantly updated it is the best SEO course that has ever been made I know I'm a little bit biased when I say that but I think you will agree if you get it it's at combat keywords.com and you're going to love it that's everything that I got for you on this episode of the show this is episode what episode is this I don't even know I'm losing track 1082 or 1083 this episode 1083 of the Edward show 1083 days in a row doing this podcast if you watch this on YouTube thank you so much for watching if you listened on Spotify or Apple podcast thank you so much for listening and I will talk to you again tomorrow bye now