AI is being pushed as a friend: your new sidekick that tackles the legwork you don’t have time for anymore. Great!
But what happens when you realize that this sidekick that helps is also scaling new ways that can directly hurt your business?
AI is both friend and foe in ways most consumers and brands are not prepared to handle.
Latest Videos FromTechRadar
Chief Marketing Officer at Forter.
Enter: AI imaging tools. These tools have become incredibly sophisticated and easily accessible, especially after recent upgrades. They can produce multiple high-quality images from a single prompt – no design background or expensive technology required.
The tools designed to help everyday users move quickly are now equally valuable to bad actors looking to make their schemes more convincing and harder to detect – and even consumers who feel pushed to abuse retail policies.
The age of AI-powered abuse is here: seeing should no longer be believing.
AI has democratized fraud
Businesses have long relied on photos and documentation to validate returns and refund requests. For a while, this logic was sound: the time, skill, and cost required to manufacture evidence acted as a barrier for most consumers, but the dam is breaking.
With modern image-generation tools, a malicious actor or your next-door neighbor doesn’t need design skills or specialist software. A single prompt can generate realistic receipts and damaged product images, turning the dial up on return abuse, refund fraud, and friendly fraud.
In fact, the Merchant Risk Council (MRC) found that over the past year, 57% of merchants reported increasing rates of refund and policy abuse. Retail’s global multi-billion-dollar fraud problem just became that much more costly, thanks to AI.
The speed and scale at which these images are produced, altered, and tested are staggering. With AI, fraudsters and abusers can now tailor claims to different merchants, test variations, and scale attacks across multiple accounts and thousands of merchants in short order.
The challenge is that AI-powered abuse is not being driven by one type of actor. Retailers are facing pressure from both organized fraud rings that deliberately exploit systems at scale and everyday consumers who are using new tools to push the boundaries of return and refund policies. While the motivations and sophistication levels are different, both create additional complexity for businesses trying to protect customers while preventing abuse.
Organized fraud groups are increasingly treating policy abuse as a scalable business model. Rather than relying on a single fraudulent claim, these groups look for weaknesses in retailer processes, create multiple accounts, and coordinate activity across merchants. AI-generated images make these operations more effective by providing convincing evidence that supports false claims, helping bad actors appear as genuine customers.
Forter has seen coordinated returns abuse operations where fraud rings used AI-altered images to support claims of product damage, including minor cracks, dents, and faulty components. In one case, Forter stopped a $1.5 million returns abuse operation over Christmas 2025 that used AI-altered damage images to make fraudulent refund requests appear legitimate. When not caught, these abusers resell the quality merchandise after getting refunded for the purchases.
The scale of these attacks is what makes them particularly challenging. A single suspicious claim may not reveal much, but when activity is connected across accounts, devices, behaviors, and merchants, a much bigger picture emerges. Fraud rings can spread activity across multiple retailers, making each individual interaction harder to identify without broader context.
AI has blurred the lines between fraud and abuse
But it’s not just organized criminals who are trying to cash in on these capabilities. The accessibility of AI image tools means everyday opportunistic shoppers are also starting to use them for their own gains. A shopper who wants to avoid the cost of a return, claim a refund for an item they damaged themselves, or take advantage of a flexible policy can now create realistic supporting evidence in seconds.
This creates a growing grey area for retailers. Not every questionable claim comes from a professional fraudster, and not every policy abuser operates with the same level of intent or organization. However, the impact is the same: businesses must spend more time and resources determining which return claims are from trusted customers and policy abusers, while legitimate customers risk facing additional friction as retailers respond to rising abuse.
Fighting AI with AI
Companies may be quick to tighten their policies to curb this AI-powered abuse. But shortening return windows, charging for returns, or delaying refunds reduces abuse at the expense of good customers. The goal has to be identifying the bad actor, not penalizing the good customer. Unfortunately, traditional measures like static rules and manual review processes can’t operate at the speed or volume AI-powered fraud now demands.
Detection must operate across the full context – identity, device behavior, transaction history, account age, claim patterns – to catch what any individual piece of evidence would miss. Machine learning is the only realistic path to doing that at scale, in real time, without creating so much friction that legitimate customers are turned away.
Visual evidence needs to be augmented with commerce context and intelligence. A damaged phone screen from a long-term customer with one prior return means something very different than a three-day-old account submitting its fifth claim this week across dozens of retailers.
Businesses need to know with confidence who is behind the claim to assess what’s legitimate and what’s not.
Check out our list of the best ecommerce platforms.
This article was produced as part of TechRadar Pro Perspectives, our channel to feature the best and brightest minds in the technology industry today.
The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/pro/perspectives-how-to-submit
