AI slop goes industrial: 354% surge and 2,100 fake recipe sites

Two independent measurements landed within a day of each other and both point the same way. DoubleVerify reports monthly ad requests tied to AI slop pages surged 354% between January and September 2026, from 184 million to 834 million, for 3.3 billion total requests in the first nine months; in a separate scan of 4.5 billion pages, 30% contained some form of AI-generated content. Separately, supply-chain intelligence firm DecryptAds found more than 2,100 AI-generated recipe sites monetized through SSPs that work with ad-tech partners in what it calls "risky geos" (Russia, China, Cyprus, Turkey, the UAE), each built with fabricated author profiles, AI avatars and even altered celebrity images. DecryptAds co-founder Zach Edwards told AdExchanger the company found no evidence of crimes, but described inventory of "dubious value" packed with placements "designed to game viewability." DoubleVerify CTO Nisim Tal framed the moment plainly: "This is a scary moment, sure, but it shouldn't be paralysing."

The crux

Generative AI has collapsed the cost of producing made-for-advertising supply to near zero, and the arbitrage that the ANA's 2023 programmatic transparency study pegged at 21% of impressions and 15% of spend is now scaling faster than inclusion lists can catch it.

Sell-side read

Every slop impression that clears a viewability floor bids against legitimate publisher inventory at a fraction of the CPM, dragging down clearing prices for the open web just as Google's own helpful-content policy starts flagging fabricated creator profiles. Publishers that sell on quality and context are competing against a machine that costs almost nothing to run.

Watch

Whether DoubleVerify and the SSPs extend AI-slop detection into a default block, and whether Google's helpful-content update is followed by a Google Ads enforcement mirror.