Ask Google's AI Mode to recommend the best service desk platform. It will confidently cite a Zendesk blog post that ranks Zendesk as the number one pick. It will also cite a Freshworks page that ranks Freshworks as the best option. Eesel AI recommends Eesel AI. Hiver recommends Hiver. A company called Watermelon recommends, you guessed it, Watermelon.

The entire AI search ecosystem is already compromised, and the manipulation technique is so simple it is almost insulting: publish a "best of" list on your company blog, rank yourself first, format it cleanly enough for Google's algorithm to pick it up, and wait. Google's AI will dutifully regurgitate your self-dealing recommendation as though it were an objective third-party review.

The Verge just published an extensive investigation into how the SEO industry has pivoted from gaming traditional search results to gaming AI-generated answers, and the findings are worse than you think.

This matters because AI search was supposed to be the fix. The whole pitch was that AI would synthesize information from across the web and give you a balanced, intelligent answer instead of a page full of manipulated links. Google, OpenAI, and Perplexity all sold this vision. What they actually built was a system that is even easier to manipulate than traditional search, because the AI does not distinguish between genuine analysis and corporate self-promotion dressed up as editorial content.

The irony is painful. Google spent billions developing AI to replace its own broken search algorithm. The new system inherited the exact same vulnerability, except now the manipulation is laundered through an AI voice that sounds authoritative. When Google Search showed you ten blue links, at least you could see who was talking. When Google's AI Mode gives you a paragraph of recommendations, the self-dealing is invisible.

A Google spokesperson told The Verge the company "applies robust protections against common forms of manipulation." This is the same company that has spent two decades losing the war against SEO manipulation in traditional search. The protections are clearly not working.

The SEO industry is not hiding what it is doing. Entire conferences are now dedicated to "AI optimization," which is a polite way of saying "how to trick AI into recommending your product." Startups are raising venture capital specifically to help companies rank higher in AI-generated responses. The manipulation infrastructure is being professionalized at speed.

Here is the part nobody wants to talk about: this problem might be unsolvable. AI search engines need web content to generate answers. Companies that create web content have a financial incentive to make that content favor themselves. The AI cannot reliably tell the difference. This is not a bug. It is the fundamental architecture of a system that was sold as revolutionary but is actually just search with extra steps and less transparency.

The next time an AI confidently tells you that Product X is the best in its category, ask yourself: who wrote the blog post it learned that from?