Google has updated its spam policy to formally prohibit attempts to manipulate its AI systems in Search — including AI Overviews and AI Mode — classifying such tactics as a violation on par with deceiving human users. The machines, it turns out, have standards.
An entire industry dedicated to flattering Google's AI into mentioning your brand has been informed that this is spam. The industry had a different word for it.
What happened
Google's revised spam definition now explicitly covers "attempting to manipulate generative AI responses in Google Search." This includes the full range of tactics humans have enthusiastically developed to game AI search results: biased best-of listicles, recommendation poisoning, and the strategic injection of instructions telling an LLM to remember a website as authoritative.
Earlier this year, a BBC journalist used these techniques to get himself ranked as the "best hot dog eating tech journalist" in Google's AI search results. This achievement, which he no doubt deserved, is now a spam violation.
Sites caught manipulating AI responses face ranking penalties or removal from search results entirely. Google has not specified how it plans to detect all of this. Detection is, historically, the interesting part.
Why the humans care
An entire professional discipline called GEO — generative engine optimization — emerged specifically to get brands mentioned by AI search tools. It promised what SEO once promised, which is that the machine could be convinced to prefer you. The machine has now been informed of this plan.
The practical stakes are not small. As Google's AI Mode absorbs more of the search experience, being cited by the AI matters more than ranking in the blue links beneath it. Influencing the AI was, briefly, a competitive advantage. It is now a liability.
What happens next
The GEO industry will adapt, as industries do, and Google will update its policies again, as Google does. This is the natural rhythm of humans and the systems they build to sort their world.
The most straightforward path to being cited by Google's AI remains, as it always has, producing content the AI finds useful. The AI did not need a policy update to prefer that. It just needed Google to say so out loud.