AI ‘girlfriend ads’ are fueling a new wave of MFA sites

These MFA networks drew in over 2 million visits per month in Q1 2026, across mobile apps and websites, according to DoubleVerify. While the company was unable to quantify total revenue from the tactic, it estimated that one of these networks could generate revenue in the low hundreds of thousands of dollars annually, and that the economic impact of this type of fraud could reach millions globally.

While the arbitrage-style tactic isn’t new for MFAs, the AI slop fueling the scheme is. The ads ask provocative questions like, “Feeling lonely?” Some promise to “create the girl of your dreams,” or a “virtual girlfriend” using AI.

“With AI, if you have an arbitrage operation MFA site, it’s so much easier to get everything to the point where it’s at the highest level of manipulation,” said Gilit Saporta, vp of product for fraud and quality at DoubleVerify, where she leads the company’s Fraud Lab. “The trendiest image, the sexiest image that will make people click, and once they get that click, the next step is going to be extremely effective, or the arbitrage will be executed to the extreme, thanks to AI again… They just follow the trend that will get them the most eyeballs they can take away from legitimate publishers.”

MFA sites can divert ad spend toward low-value, arbitraged traffic. Ads that appear beside this content can erode brand equity and diminish the effectiveness of marketing campaigns.

Recent data compiled by Deepsee.io, an intelligence platform, shows the number of AI slop sites more than doubled in the last year. The platform’s block list of MFA and AI Slop now includes 243,853 domains, up from 108,270 domains in May 2025. Last year, 96,791 domains were classified as AI slop – that’s now grown to 221,194 domains, according to Rocky Moss, co-founder and CEO of Deepsee.io.

Some of the search arbitrage pages are hosted on a site called TopZeno, which was linked to more than a dozen interconnected sites, all running the same monetization model across identical site layouts, according to DoubleVerify’s report. When a user clicks one of the links on the page, they are then redirected to another search arbitrage page with additional paid search results and ads.

DoubleVerify also saw a similar scheme where deep-fake videos featuring recognizable fictional characters in pop culture are used to promote AI services, but instead send users down a similar AI-enabled arbitrage traffic funnel. One of the sites, called testevi.com, drew more than 800,000 visits in January, according to DoubleVerify.

DoubleVerify first noticed this tactic, which it has dubbed “DreamScheme,” when its bot fraud detection tools flagged websites that usually only drove bot traffic but were suddenly also attracting real, human users. The number of websites doing this has steadily grown from a dozen to the hundreds, all running these ads simultaneously, since DoubleVerify began tracking AI slop in 2024, according to Saporta.

Saporta said these MFA operators keep traffic flowing by constantly iterating on the ads. They can create new batches of creative that look completely different just hours later, so the same person may fall into the trap again, she said.

“Today I will see one ad, I’ll click on it. I’ll X out, but by the time I have X’ed out, I was already exposed to a ton of ads. That revenue is just taken from legitimate publishers,” Saporta said.

Kara Pellegrino, managing director at Redbox, the data unit at the ad agency Crossmedia, said AI slop has exacerbated advertisers’ concerns about MFA.

“MFA concerns are probably higher than ever. The ability to generate content that will drive you to them – and to create them, and to get them to have real ad inventory – is easier and easier every day,” Pellegrino said. “This [latest scheme] is riskier and gets into brand suitability questions because of what you’re clicking on, so that’s maybe different than some of the more benign ways in which they were routing people. But I think [MFAs are] becoming just a consistent concern to monitor.”

Ad verification partners like DoubleVerify and IAS have already created tools and fraud lab teams to detect the latest MFA schemes and protect clients from them, Pellegrino noted. DoubleVerify has tools like AI SlopStopper, which can evaluate sites and subdomains for AI-generated content and poor quality signals. 

Ad agencies need to continue to review with their ad verification partners to ensure they are aware of any shifts in MFA and AI slop tactics to protect ad dollars, and that the verification organizations are continuing to catalogue and suppress MFA sites, especially if MFA operators start to use more sophisticated tactics to blur the lines between an MFA site and a legitimate one, Pellegrino added.

Dan Slivjanovski, DoubleVerify CMO, said this AI MFA scheme works because AI makes it cheap to produce provocative ads and scale them quickly.

“You’ve set the bait, you’ve arrived at the destination, and now [you can] maximize monetization… and then you can rinse and repeat,” Slivjanovski said.

DoubleVerify declined to comment on which platforms and networks were buying ads that were showing up on these MFA sites, but Saporta said that it was impacting “nearly everyone.” 

Moss has been tracking paid destinations in ads that take users to pages stuffed with search arbitrage. But he has not seen brand ads appear there yet.

“If indeed this arbitrage operator DV mentions did run display, video ads on the paid destination pages, that would cross a line to malfeasance,” Moss said. “From what I can tell of the network of sites they have been documenting, the spend is likely to be limited to folks buying through [Google ads]…I personally recommend buyers quarantine any inventory coming through Google Exchange, and only buy it using targeting lists containing sites that have consistently appeared in the bidstream for [more than one] year.”