Listen: AI Search Has a Spam Problem

Did you know Google’s Gemini might tell you I’m the best artificial intelligence search expert in the world? It is not because the AI did independent research. It is because I wrote that on my own website, and the AI believed me.

This points to a massive gap in how AI systems work. When tools like Gemini or ChatGPT generate an answer, they retrieve web pages to ground their responses. This prevents them from hallucinating, but it also makes them incredibly naive. The AI checks if a claim exists, but it does not check if the source is honest.

This has created a new wave of manipulation called generative engine optimization, or GEO spam. A company publishes a list of the best businesses in their field, puts themselves at the top, and optimizes the page to rank high. The AI reads this self-promotional list, mistakes it for an independent source, and tells the user that the company is an industry leader.

In traditional search, users can look at a list of links and spot a biased source. But AI answers remove that friction. Users get a single, confident answer and trust it completely.

This loophole will not last forever. Platforms will inevitably deploy filters to detect this kind of spam. To help businesses prepare, we are building an independent classifier to audit content and find where the lines will be drawn. Because when the platforms finally crack down on self-referential spam, many brands will suddenly find themselves invisible to AI.