Scaled AI content is the practice of publishing large volumes of AI-generated pages faster than any human review could vouch for them, in order to capture search or AI-answer traffic. Google names the pattern directly in its spam policies as scaled content abuse: generating many pages primarily to manipulate rankings, regardless of whether a human or a model wrote them. The policy is about the intent and the volume, not the tool.
The Mount AI shape
Glenn Gabe coined the term Mount AI for what these programs look like on a traffic chart: steep growth followed by a similarly shaped drop once Google’s systems have gathered enough signals to identify what is going on. Lily Ray then documented it at dataset scale. Across 220 or more sites drawn from AI content platform customer lists, 54 percent lost 30 percent or more of their peak organic traffic, 39 percent lost half, and 22 percent lost three quarters. The typical timeline was six to twelve months of page growth, a traffic peak three to six months later, then a steep decline within the following year.
It does not stay a Google problem
AI answer engines retrieve from search indexes, so the crash travels. In Gabe’s case study of a site that received a manual action for scaled content abuse over roughly 850,000 AI-generated pages, the site’s citations in ChatGPT fell away alongside its Google visibility. Being cited in AI answers depends on staying indexed and trusted in the sources those answers are built from.
How Citate treats it
Citate will not generate a page it cannot ground. Every draft is built from the client’s own knowledge base of grounded atoms, and the penalty gate reads a site’s URL structure against the live spam policy to flag clusters that have the scaled shape before they are published. Volume is never the goal; the answer changing is.