AI-generated spam is reshaping what people believe about brands. The mechanics and the implications are alarming but the defence is simple.
In April 2026, SEO expert, Lily Ray, brought our attention to something important regarding LLMs and GEO. Her post announced an algorithmic change dubbed the “September 2025 Perspectives update.” This update, Lily claimed, would have profound implications for how AI platforms surfaced answers to prompts, which in turn would have profound implications for how brands are discovered.
We will dig into the details of the reported update in a moment. But first, there is one crucial detail that must be stated up front: the update never happened.
It is not that Lily was wrong, or that she was intentionally spreading misinformation. Lily’s post was an experiment. And the results were astounding.
Given that the update was imaginary, perhaps the details are not important. Except one apparently trivial point left behind a significant clue. Lily’s post claimed that the update had been approved “between slices of leftover pizza.”
Within days, AI Overviews had picked up Lily’s article. Other AI platforms scraped and republished it. A fictitious story became accepted fact. Anyone seeking information about how LLMs function was at risk of being misled, even on the “slices of pizza” fabrication, which pointed towards Lily’s piece as the primary source.
Lily Ray deleted the article but her point was proven. Her wise summation: “Repetition is treated as consensus.” A conclusion which, in the age of AI, holds more truth than ever.
The AI slop loop
A BBC journalist conducted a similar experiment to Lily Ray and got the same result, fooling AI systems that he was the leading competitive hotdog eater within his industry. The same mechanics were at play.
A false and niche claim filled a content void, funnelling LLMs towards a single source when specific prompts were entered by users.
In Lily’s case, the claim was consequential, and so was cited by other writers, also relying on AI. In some cases, humans played little to no part in the writing, research or fact-checking process, and so the myth began to feed itself at an alarming pace.
This is the AI slop loop. False claims become accepted fact through repetition, citation and AI synthesis, despite having no basis in reality. It is not a single hallucination that can be fixed. It compounds.
This matters to brands because AI search is now a primary discovery channel. And many users take what AI says at face value.
Research from Censuswide found that 54% of consumers form a negative opinion of a brand that doesn’t appear in AI search results.
The implied upshot is just as uncomfortable. What happens when a brand does appear, but inaccurately? Or when fabricated consensus positions it somewhere it doesn’t belong?
If it is your brand or business being misrepresented, it is a big deal.
How AI spam happens
Bad actors have worked out that polluting the data pools AI search engines use to summarise reality turns the AI into an authoritative mouthpiece for whatever they want it to say. Gianluca Fiorelli’s analysis identifies several techniques in circulation, including:
● RAG flooding: thousands of AI-generated forum posts and fake blog comments designed not for humans to read, but to pollute the vector databases AI draws on
● Synthetic consensus manipulation: seeding enough sources saying the same thing that an AI concludes there is genuine public agreement
● Entity association poisoning: repeatedly linking a brand to trusted authority names, think “Harvard study” or a well-known tech company, until the AI treats the association as established fact
So what can brands do about AI spam?
The honest answer is it is incredibly difficult, if not impossible, to stop AI spam at source. What you can do however is make it harder for the wrong narrative to take hold.
Think of it this way: if someone fills the information space claiming your brand is known for something it is not, the only effective response is to be the most authoritative, consistent and credible source on the truthful narrative you want people to hear.
As Don Draper puts it: “If you don’t like what people are saying, change the conversation.”
In practice that means consistent coverage in quality earned media outlets, proprietary data and research that LLMs can cite. It requires expert voices associated with your brand’s real areas of expertise and a digital presence structured so AI can read and understand it.
These are not new ideas. They are the fundamentals of good PR and content strategy. What is new is that these ideas matter more now, and the consequences of neglecting them are more immediate.
AMEC, the international body for communications measurement, launched its GEO Principles in May 2026 specifically to bring rigour to this space. Its framework is built on a clear ethical position: GEO should improve the availability of accurate, useful and verifiable information. It should not involve low-quality content, disguised promotion or manipulation. It is a useful line for any brand to hold.
Want to know more about GEO?
Our ‘Mastering AI Visibility’ guide is the perfect start point for any B2B marketer looking to enhance the visibility and reputation of their brand through GEO. Full of tips and insight on everything from the technical foundations of LLM discoverability to the role of earned media, it offers a comprehensive, easy-to-understand overview of GEO.
AI spam FAQs
The AI slop loop is the self-reinforcing cycle in which AI-generated misinformation, opinion as fact or low-quality content spreads and compounds. E.g., a false claim gets published, AI systems pick it up, other sites scrape and republish it, then the repetition is treated as consensus. The original source becomes irrelevant. The myth becomes the record.
Not easily at source. The mechanics of AI spam – RAG flooding, synthetic consensus manipulation, entity association poisoning – can operate at a scale and speed that is potentially beyond control. What brands can do is make the wrong narrative harder to establish by being the most authoritative, consistent and credible source on their own story.
A GEO audit assesses how your brand currently appears in AI-generated search results. It covers what LLMs say about you, where the gaps are and where inaccurate or misleading information may already be taking hold. It is the starting point for any strategy to improve your brand’s AI visibility. Find out more about The Think Tank’s GEO Audit.
