By Gilit Saporta, DV VP Product, Fraud & Quality
Excerpt from the forthcoming The Fraud Fighter's AI Playbook, by Gilit Saporta, Chen Zamir and Shoshana Marany. Sponsored by the DV Fraud Lab.
Generative AI is fundamentally transforming the volume, quality and tactics of online fraud. DV fights back with transparency.
The Fraud Fighter's AI Playbook sheds light on AI-powered fraud schemes, providing a comprehensive framework for countering these threats, from updated detection methodologies to accessible protection strategies. This excerpt is adapted from Chapter 3.
Deception and evasion tactics were here long before GenAI. It’s just that now they happen at scale. These deception methods impact identity verification and trust; i.e. “How do I know that I’m transacting with a genuine, trustworthy human being?”
These examples from three different industries reveal the scope of this threat.
In 2025, a website called "Dorothy Bags" announced that after 35 years, Dorothy, a UK artisan, was selling her final collection of quilted leather cross-body bags, which apparently over 10,000 Britons cherished and loved, for 80% off. The announcement came in the form of a bylined, reposted “Artist of the Week” article from a news site called Times Weekly.
But there is no Dorothy. The images and bio were made with GenAI, with minimal human oversight. The contact details provided for Dorothy’s are not found on any map. The headshot for the “reporter” who covered the “Artist of the Week” belongs to a real reporter, but the website gave him a new name.
Last but not least, the “handmade leather bags” listed on the website are not UK-made, nor handmade, nor leather. They can easily be found via image search, sourced from low-cost marketplaces in Asia. Dorothy Bags likely supports a drop-shipping operation. It uses an AI-generated story to resell merchandise at a handsome margin. Dozens of replicas of this site, offering various items from other “artisans” (Jean's jewelry, Vera’s lingerie, Amar’s dresses), target shoppers in Europe, Asia and North America. If you work with sellers or as a marketplace, when customers have complaints about the product or experience, they’ll be coming to you.
Investment scams are a step more pernicious than eCommerce scams. Deepfakes and GenAI content have been powering pump-and-dump scams, cryptocurrency scams, job hiring scams and so on. Notoriously, GenAI-powered schemes hijack trusted public figures. In one instance, Martin Wolf, a reporter for the reputable Financial Times, found that a deepfake avatar of himself was being used to draw victims into investment scams. Imagine building a respected reputation for decades, only for a fraudster to undo all of that goodwill.
GenAI tampers with our human intuition and judgment of character. And the financial impact is staggering: Deloitte estimates that AI-powered fintech fraud will reach $40 billion by 2027.
When it comes to financial impact, adtech fraud schemes dance circles around fintech and eCommerce fraud: Juniper Research projects adtech fraud losses will top $170 billion by 2028. Because adtech is complex and involves many intermediaries, it only takes one bad middleman to thwart the supply chain. Resellers can unwittingly participate in a fraud operation, but the overall damage of fake or fraudulent media (or fake/manipulated/bot traffic) spreads through two to three platforms. This magnitude of losses makes adtech fraud the second most lucrative form of crime, surpassed only by drug trafficking, according to the World Federation of Advertisers.
The following example shows how simple it is to make yourself look like an adtech reseller. In its simplest form, the fraud works like this: Fraudsters create a fake company and position it as a legitimate ad reseller. The company collects real advertising dollars and then delivers those ads to fabricated websites or bot traffic. No real person ever sees the ads, but the advertiser still pays.
This bogus site, “Unholy Tech Limited,” managed to act as a media reseller for months, selling successfully to a few respectable supply-side platforms. To add credibility, the website includes a “meet our team” section. However, the names are made up and the photos are stock images.
At the DV Fraud Lab, we identify these schemes through pattern recognition and behavioral analysis. "Dorothy Bags" revealed multiple red flags: mismatched contact information, recycled AI-generated content across domains, and product images traceable to low-cost marketplaces. Fintech deepfakes exhibit inconsistent metadata and distribution through suspicious social ad networks. And finally, adtech fraud shows telltale signs of templated websites, fabricated testimonials and traffic patterns inconsistent with legitimate publishers.
Trust and transparency is the bedrock of the media transaction that powers the free internet we all consume and enjoy. These schemes threaten trust across the digital ecosystem. When fraudsters use GenAI to fabricate identities, they don't just steal from individual consumers, they poison the well for legitimate businesses. Publishers, advertisers, agencies and platforms all need to double down on vetting efforts. The line between harmless misrepresentation and outright fraud has never been blurrier and the scale, speed and accessibility of GenAI-powered deception demands urgent action.
GenAI makes deception easier, but detection is evolving too. In The Fraud Fighter's AI Playbook, my co-authors and I provide a comprehensive framework for countering these threats, from updated detection methodologies to accessible protection strategies. Whether you're leading fraud operations or strengthening your defenses, this playbook will help you stay ahead.
The Fraud Fighter’s AI Playbook will be available in November. A free sample chapter is available for download now.