Websites filled with machine-generated filler have increased fivefold, according to a report Axios published this week, and the driver is a new optimization practice the trade calls GEO — generative engine optimization.

That single number is the story. Five times more low-quality AI content sites now compete for attention inside the answers that chatbots and AI search tools hand to users. Axios framed the growth around GEO, the practice of tuning pages so that AI systems, rather than classic Google rankings, pick them up and cite them.

What is GEO, and why does it matter here?

For two decades, publishers optimized for PageRank: keywords, backlinks, page speed. GEO targets a different judge. Instead of ranking a list of blue links, an AI assistant synthesizes an answer and pulls from whatever sources it judges relevant. A page that wins that sweepstakes gets quoted directly in front of the user.

The Axios report points to an obvious consequence. If AI systems reward content that reads authoritative, then mass-produced text becomes a product you can manufacture cheaply and point at that reward. The fivefold figure suggests plenty of operators have done exactly that.

A caveat belongs here, and it applies to the whole story: the report gives a growth multiple, not a base count. Five times a small number is still a small number; five times a large one is an industry. Axios did not publish the underlying volume in the headline finding, so the scale of the problem remains an open question until someone counts the sites directly.

Who benefits from the flood?

The economics are simple and not new. Generated text costs almost nothing to produce at volume. If even a fraction of slop pages get cited by an AI answer engine, the operator collects traffic, ad impressions, or affiliate clicks that cost them pennies to acquire.

The losers are equally easy to name:

• Original publishers, whose reporting AI systems may paraphrase while citing a copycat page
• Readers, who cannot easily see which source behind an AI answer did actual reporting
• The AI platforms themselves, whose answers inherit whatever quality their sources carry

Axios did not name specific slop operators or platforms in the headline finding, so assigning blame beyond the pattern itself would go past what the report supports.

Is this just SEO spam again?

The pattern rhymes with history. Content farms plagued Google search in the early 2010s until algorithm updates cut their revenue. GEO optimization looks like the same arms race moved one layer up the stack — from gaming a ranking algorithm to gaming a summarizer.

There is one difference worth noting. A search engine shows users ten links, and users can skim past a bad one. An AI answer shows one synthesized paragraph. If the paragraph leans on a junk source, the reader often never sees the alternative. That concentrates the cost of a bad citation.

Whether AI vendors can build defenses as effective as Google's panda-era updates remains unproven. No platform has published numbers on how much slop its systems currently cite, and without that baseline, the fivefold growth figure cannot be checked against actual answer quality.

What should operators and readers take from it?

For publishers, the report is a warning about measurement. Traffic dashboards built on search referrals will miss the AI-citation channel entirely. Publishers do not yet have a standard audit tool for knowing how often AI engines quote them versus quote a scraper.

For readers, the practical rule is old: check the source behind the answer. That is harder when the answer blends ten sources into one paragraph, which is precisely the problem.

Axios has put a number on the growth of AI slop sites. Nobody has yet put a number on how much of it reaches real users through AI answers — and that is the figure that will decide whether this is a nuisance or a crisis.

Fivefold growth in supply means nothing until someone measures the demand.