Generative AI content farms are stealing publishers' ads.txt files to hijack ad revenue, according to DoubleVerify, which has been tracking the activity since January.
The tech company has been investigating what it has dubbed "AI slop sites and networks" - one in particular called "Synthetic Echo" - which uses generative AI to mass-produce content and spoof ads.txt files to siphon revenue from legitimate publishers.
These sites are populated with low-quality, AI-generated content scraped from other reputable publisher sites like headlines, images and layout, to recreate the site on a fake domain that looks nearly identical, to drive ad revenue.
DV has uncovered URLs like https://nbcsportz.com/, espn24.co.uk, cbsnews2.com and 247bbcnews.com which have been designed to mislead ad tech vendors and buyers by mimicking reputable publishers by scraping their content. Ads have been running on these sites for months.
Now fraudsters are getting even bolder in their tactics, according to DV's latest findings -- copying and pasting entire ads.txt entries onto their pages to make it look even more like they're trusted publishers, to major ad-buying platforms. It has uncovered 100 sites like this so far.
The tech company is issuing its latest findings in a seven-page document guiding publishers and marketers on how to address the problem, which includes tips on how publishers should continue to actively audit their ads.txt files and ensure they remain tight with only trusted sellers.
It's a familiar story with a new twist. AI may be the shiny new layer, but beneath it lies the same old ad tech problem: fraud in fresh packaging.
Gilit Saporta, vp of product management at DoubleVerify's Fraud Lab, who has been tracking digital fraud for decades, said that generative AI is amplifying the existing issues. "It's ad tech fraud schemes but on steroids, because gen AI lets you amplify and replicate a bad website that has been set up either to spread fake news, hate speech, or just to monetize it for the sheer traffic that you can sell on this side, whether it's real traffic or just bot traffic," she said.
In other words, generative AI hasn't created new types of ad tech fraud -- it's just made the old ones easier to execute and scale faster than ever. "With gen AI it's so much easier to make the site look plausible and replicate the site to 1,000 or 10,000 sites a day," added Saporta. "We're not necessarily seeing an increase in the sophistication of schemes this year, but we're seeing an immense uplift in the scale," she said.
"It's extremely disturbing that premium news publishers - once again - are being abused by fraudulent players in the ad ecosystem," said Thomas Lue Lytzen, director of ad sales and tech at Danish tabloid Ekstra Bladet. "It only makes it even more relevant for advertisers to work with whitelists of quality publishers and not just letting the buying algorithms pick and choose."
Ads.txt was introduced by the IAB Tech Lab in 2017 as a potential fix to help buyers verify which publishers are authorized to sell inventory. While it's no silver bullet to take down bad tactics, it has helped many publishers to shave off unauthorized resellers, in the hope that more of the ad buys will go to the publishers, rather than go to waste.
Programmatic has always been a game of whack-a-mole and this is just the game moving on, stressed Hamish Nicklin, who during his tenure as CRO of Guardian News and Media became notably vocal against ad tech malpractices.
When describing the latest issue with what DV has called ads.txt plagiarism, he said: "Let's be clear: this isn't an AI problem, it's an ad tech problem. The pipes are still too easy to spoof. If advertisers want to chase the cheapest CPMs across the open web, they have to accept the risk that comes with it," he added.
He believes that the real solution is for buyers to spend in safe, curated marketplaces and rebuild direct ties with premium publishers, while continuing to hold ad tech vendors to account for who gets to be in the exchanges in the first place.
Other publishers mirror the same sentiment: that there will always be new issues that plague the open auction which are near impossible to solve. "There will always be new issues that are gamed by fraudsters," said Pritesh Jumani, head of ad tech and performance media and revenue at Globe and Mail.
Naturally, publishers are seizing the moment to tout direct deals. Unlike the murky waters of the open auction, where spoofed inventory and slop sites thrive, working directly with publishers via private auctions, preferred deals, or programmatic guaranteed not only mitigates risk but also offers buyers greater transparency, stressed Pritesh.
For years, the ad tech ecosystem has been riddled with inefficiencies and bad actors -- from opaque vendor fees to fraudulent inventory that siphons ad dollars away from real journalism.
Nevertheless, the rise of AI-generated content farms plagiarizing legitimate publishers' ads.txt files marks a troubling new chapter in the long-running story of digital ad fraud.
Danielle Coffey, CEO of News/Media Alliance and Richard Reeves, managing director of the Association of Online Publishers in the UK, both expressed "significant concern" at the latest DoubleVerify findings. And both stressed their intention to monitor the issue closely, and work with industry shareholders, including IAB Labs, to address the issue. IAB Tech Labs didn't respond in time for the publication of this article.
Ultimately, it's nothing new for publishers.
"It seems to be in the interest of most ad tech companies involved in programmatic to keep the show going exactly as it is," said Alessandro de Zanche, independent media consultant and former News UK director of programmatic. "We clearly have a problem: digital advertising's sink is overflowing. But instead of fixing the tap, the industry is gathered around the sink debating whose mop is best."
The reason has been the industry's most open secret for some years. "If we fixed the tap, the business models of most ad tech companies would collapse," said Zanche.
Instead, he believes the necessary clean-up step is to establish advertising environments where publishers, advertisers and agencies and those representing them, are legally identified as a prerequisite to trading. "Then we can start talking about standards and keep chopping off the dead branches. But we don't need AI for that, we need a clear will, starting from SSPs and DSPs," said de Zanche.