Back to All Posts

AI Slop and MFA Sites: A Buyer's Checklist

DecryptAds mapped 2,116 AI-generated recipe sites carrying major brands' ads. What Google and YouTube changed in October 2026, plus a buyer's checklist.

AI slop and MFA sites: a buyer's checklist, Black C Media

AI slop sites are web pages mass-produced with generative AI, fronted by fabricated authors and built to sell ad impressions rather than to serve readers. A DecryptAds report published on 18 October 2026 mapped 2,116 of them in the recipe niche alone. This article explains how they reach advertiser budgets, what Google and YouTube changed in October 2026, and what a media buyer in Türkiye can check before the next campaign.

What is an AI slop site, and how does it differ from MFA?

A made-for-advertising (MFA) site is a website whose primary purpose is to sell ad impressions, not to serve a reader. An AI slop site is the current form of MFA: text generated at scale, AI-generated author headshots and invented biographies that make the site look like the work of a real person.

DecryptAds counted a site as a confirmed AI recipe site only when it showed both AI author photos and AI text indicators, checked manually on at least three pages. From roughly 400 confirmed sites, more than 190 technical fingerprints (ads.txt account IDs, DNS and WHOIS data, page scan signals) led to 2,116 domains run by the same operator clusters. The analysis ran from July to October 2026.

The monetisation pattern is the part that matters for buyers. Most of these sites outsource their ads.txt file to a sales house: 77.1% declared one Italian manager domain and 19.2% a second, US-based one. The publisher is paid by the sales house, not by the ad systems listed in the file, and often does not know which of those systems receives visitor data on every page load.

Diagram: how an AI slop site is monetised through an outsourced sales house and dozens of ad systems
How budget reaches an AI slop site: the publisher sees the sales house, not the ad systems behind it.

How much ad budget goes to made-for-advertising sites?

The share has fallen since 2023 but has not disappeared. Figures from three independent measurements:

MeasurementFindingPeriod
ANA programmatic transparency studyMFA sites took 21% of impressions and 15% of spend among the 21 member companies studiedSept 2022 to Jan 2023
Pixalate MFA benchmarkMFA domains received 10% of global open programmatic web spend; 10% in EMEAQ1 2025
IAS Media Quality Report, 21st editionMobile web display is 45.1% of measured impressions but 71.9% of MFA impressions; MFA rate 2.0% on mobile web against 0.5% on desktop2025 data, published July 2026

The practical reading: the problem concentrates on mobile web display, and a buyer who tightens that channel removes most of the exposure.

Why does Google's October 2026 update matter?

On 1 October 2026 Google added a paragraph to its helpful content documentation: fabricating creator profiles with AI-generated headshots, made-up names or false credentials is a form of deception, and deception is a signal of a low-quality page. The Search Quality Rater Guidelines already instructed raters to give such pages the Lowest rating. AI-assisted content itself remains eligible for Search; the test is honest authorship and usefulness, not the production method.

For advertisers this is a usable standard. If Google treats an undisclosed AI persona as a mark of low quality, a media plan can apply the same rule to inventory.

What changed on YouTube?

YouTube replaced its "repetitious content" rule with an "inauthentic content" rule in its Partner Program and listed content it will not monetise, including AI characters that give advice on finance, health and similar topics. A Kapwing study from January 2026 found that 16 of the 100 most-subscribed AI slop channels had been removed from the platform. In a separate Kapwing test, 104 of the first 500 Shorts served to a fresh account were entirely AI-generated.

In Türkiye, where YouTube is the largest media platform by reach, this is not a side issue. The platform's own enforcement reduces the volume, but it does not decide suitability for an individual brand.

Is all AI content a brand risk?

No. In March 2026 Zefr and OM Media Trials published the first study to measure ad performance next to different types of AI-generated content, with a consumer panel in the United States and Canada. The main findings:

  • 32% of respondents misidentified human-made content as AI-generated.
  • 81% said there is at least one type of AI content a brand should not appear next to.
  • 41% felt more positive about a brand when the AI content was clearly labelled.
  • When viewers could not tell whether content was AI or human, brand favourability, trust and purchase intent all declined.
  • Satire, humorous youth depictions and creative expression lifted ad recall and perceived innovation.

The conclusion is consistent with Google's rule: the damage comes from deception and ambiguity, not from the tool. The panel is North American; a Türkiye panel does not exist yet.

Diagram: AI content sorted by disclosure and quality, from creative and labelled to fabricated and undisclosed
Two axes decide suitability: is the content disclosed, and is it made for a reader or for an ad slot?

A note on "risky geo" lists

DecryptAds built its sample by starting from ad systems registered in jurisdictions it classifies as risky: 81 in Russia, 85 in China, 26 in secrecy havens such as Cyprus, and 104 in an "elevated risk" tier that includes the United Arab Emirates, Türkiye, Pakistan, Nigeria and Ukraine. A country label is a screening shortcut, not evidence about any company's content standards. One sales house quoted by AdExchanger made the same point about its Turkish partners: a NATO member, not under sanctions, with GDPR-aligned contractual obligations.

The useful test for a buyer is behaviour: does the inventory disclose its authorship, does the page serve a reader, and does the supply chain know its own partners.

A buyer's checklist for Türkiye and the region

  1. Authorship. Verifiable bylines, or a visible AI disclosure. Undisclosed AI personas are treated as low quality, following Google's own standard.
  2. ads.txt and sellers.json. Who is the declared manager domain, how many ad systems does the file authorise, and can the publisher name them.
  3. Ad density. Overlapping units and ads covering the author photo are MFA markers, whatever the content says.
  4. Mobile web display. The channel that carries 71.9% of MFA impressions in the IAS data gets the tightest inclusion lists.
  5. Video. Classification at content level, not channel level, with AI thresholds set by sub-type: creative and labelled content allowed, spam and fabricated personas excluded.
  6. Measure the change. Run a line-item holdout and report relative differences, including cost metrics, before scaling.

Key points

  • DecryptAds mapped 2,116 AI-generated recipe sites monetised through outsourced sales houses, with ads from major brands on them.
  • Google now classifies fabricated creator profiles as deception and a low-quality signal; YouTube does not monetise inauthentic content.
  • Zefr's research shows the brand risk comes from deception and ambiguity, not from AI content as a category.
  • Mobile web display carries most MFA exposure; authorship and ads.txt checks remove most of the rest.

Sources: DecryptAds, AI Recipe Slop Networks, 18 October 2026; AdExchanger, 20 October 2026; Search Engine Journal on Google's 1 October 2026 guidance; Zefr and OM Media Trials, AI Slop or Not?, 2 March 2026; Tubefilter on Kapwing's January 2026 study; Pixalate, Q1 2025 Global MFA Benchmarks Reports; ANA Programmatic Media Supply Chain Transparency Study, December 2023; IAS Media Quality Report, 21st edition, July 2026.

Related: Black C Media becomes Zefr's official partner for Türkiye · Zefr on YouTube.

Understanding, Explaining, and Advancing Advertising Technologies in MENA

We explore how advertising technologies work in MENA, how they are measured, and how they generate real-world outcomes.As C Group, our goal is not to make technology more complex, but to make it meaningful, accessible, and actionable. Because the future of media is built on measurable intelligence.
‍
Our vision at C Group is to contribute to a more transparent, more measurable, and more effective media ecosystem by bridging global AdTech standards with local market needs.