DecryptAds identifies 2,116 AI recipe sites as ad-tech firms dispute its findings
The study links ad-heavy recipe pages to advertising intermediaries. Company responses expose a key limitation: a technical connection does not always mean active monetization.
In an investigation released October 20, DecryptAds mapped 2,116 AI-generated recipe sites using ad-tech data, domain records and more than 190 technical fingerprints. The findings expose a large made-for-advertising network, but they do not establish that every listed intermediary directly monetized the sites: public records can be stale or wrong, and several firms dispute the attributions or the assumption that AI-written content is inherently low-value. DecryptAds said it found no evidence of crimes, leaving advertisers with a verification problem rather than a proven misconduct case.
01
DecryptAds began with about 300 ad-tech companies and narrowed food-keyword results to 400 suspicious recipe sites before expanding the network.
02
It attributed monetization of 1,631 sites to HBAgency, which said it works directly with only Monetix Ads among eight partners cited in one example.
03
MonetizeMore said it was not actively running advertising demand on any of 11 sites attributed to it; Bakerias.com had a legacy reporting connection.
DecryptAds says it has identified 2,116 AI-generated recipe sites packed with ads, with major brands appearing on their pages, according to AdExchanger’s coverage. The study, released October 20, traces connections to advertising intermediaries—but several named companies dispute or qualify those links, exposing how difficult it is to establish who actually helps these sites earn money.
The sites combine low-effort recipes with overlapping ad placements, following the made-for-advertising model: pages built around advertising revenue. DecryptAds research chief Zach Edwards stressed that the investigation found no evidence of crimes by the sites or their advertising partners.
From 400 suspicious sites to a larger network
DecryptAds began with about 300 ad-tech companies in jurisdictions it classifies as risky. Food-related keywords narrowed their connected publishers to 400 suspicious recipe sites. Edwards said he manually checked each for AI-generated avatars and concealed authorship. The team developed more than 190 technical fingerprints to find other sites operated by the same owners.
Its tracing combined public advertising files called ads.txt and sellers.json with domain-name records, security certificates and Google Analytics data. But those public files can contain errors and go without updates, complicating efforts to establish advertising relationships.
The report screenshot shows an Add Cakes author profile with a woman cooking and text describing “Elena.”Source: adexchanger.com.
A listed connection is not always a paying relationship
DecryptAds attributed monetization of 1,631 sites to HBAgency, a supply-side platform, or SSP, which helps publishers sell ad space. HBAgency said it works directly only with Monetix Ads among eight partners highlighted in one example. It described AdMatic as Monetix’s technology provider, not its own direct commercial partner.
HBAgency said every publisher is screened by an independent fraud provider. It also described invalid-traffic checks, protection against malicious advertising and ongoing publisher reviews.
MonetizeMore CEO Kean Graham said MonetizeMore was not actively running its own advertising demand or its partners’ demand on any of the 11 sites attributed to it. One, Bakerias.com, retained a legacy reporting-platform connection but used its own AdSense account. Graham said it had failed MonetizeMore’s traffic and advertising-quality requirements. The company initiated a wider inventory review.
The disagreement over what deserves advertising
Graham said advertisers should not buy inventory relying on deceptive authorship, and that MonetizeMore does not monetize it. Ezoic, which DecryptAds linked to 407 sites, challenged a different premise: that AI authorship inherently makes content low-value. It argued that content should be assessed for quality and usefulness rather than how it was produced.
Ezoic said it rejects more than 98% of applicant sites, combining automated checks with human review. Its assessments consider content, audience engagement and brand safety.
Sources
adexchanger.comAI Slop Sites Are On The Rise. Is Ad Tech Ready For Its New MFA Challenge?
Reader comments
Newest comments first. Replies stay oldest first.