AI’s expanding role in ecommerce comes as no surprise to either consumers or retailers. Indeed, the rapidly changing landscape of ecommerce tech has brought newfound efficiency and convenience across both operational and customer-facing processes.
The downside? The rise of a new class of sophisticated fraud risks.
On one hand, customer experience has never been more focused on convenience than today, and AI agents have supercharged the ecommerce landscape from tip to tail. Consumers can now find the right products at the best prices faster and more seamlessly than ever before.On the other hand, the use of AI causes a particular challenge when it comes to fraud, because it makes it difficult to differentiate between genuine customers and scammers. As retailers adopt more and more products that put nearly full control of the shopping journey in the hands of AI agents, new questions arise around fraud and how the retail sector can level the playing field.
Fortunately (and somewhat ironically) AI itself offers the most effective defense against these evolving fraud threats. With AI increasingly powering ecommerce and evolving at breakneck speed, retailers must harness this technology not only to improve efficiency and automation, but as a critical tool in fraud prevention.
The New Fraud Risks
New research highlights the stark and growing risk of fraud that is afflicting the ecommerce sector today.
Ecommerce fraud globally totalled $37bn (£27bn) in 2024, with projections anticipating that figure will balloon to a staggering $100bn (£74bn) by 2029. Meanwhile, false declines – legitimate transactions that are mistakenly rejected by a merchant or faulty fraud prevention system – cost more than $308bn (£227.32) in global revenue lost annually. The AITE Group predicts that merchants could lose up to 75 times more revenue to false declines than to actual fraud.
Additionally, returns abuse, where shoppers exploit a business’s return policy or manipulate refund scenarios for personal gain, was estimated to cost $103bn in the US alone last year. A recent report showed 84% of merchants admit that identifying fraudulent activity has become harder, while 76% say policy abuse is growing increasingly sophisticated.
This problem is multifaceted and costly, and it’s only getting harder to weed out the bad actors – especially when fraudsters can use AI tools, like Google’s shopping assistant, to purchase items digitally, without ever coming into contact with the seller directly.
Evolving Fraud Methods
Consumers are using AI to research faster and enhance their shopping experience. Merchants are using it to improve operations and safeguard their bottom lines.
Fraudsters are happy to capitalize on both scenarios.
Bad actors see customer and retailer touch points as two entries to exploitation, harnessing AI to undermine essential trust between retail merchants and customers and attach risk to an increasingly popular use of ecommerce.
Fraudsters now collaborate in active online communities on forums and chat platforms, such as Reddit, Telegram, and the dark web; they share advice about fraud, and even sell ‘fraud-as-a-service’ type kits, with the expressed aim to exploit merchants’ terms and conditions – the very frameworks intended to protect merchants and honest customers alike.
Even beyond crowdsourced schemes, new AI tools like WormGPT are making it easier to automate attacks, and scale engineering scams. Now, even the most rookie fraudster can rack up thousands in loot in minutes.
How Can Merchants Fight Back?
Merchants who fail to adapt to this new AI-driven wave of attacks, still relying on manual reviews, rigid return policies, or outdated fraud models will suffer on two fronts: Firstly, they will experience losses and take a hit to their bottom line; second, and perhaps more significantly, those outdated methods of fighting fraud will be less effective, and could punish loyal customers.
Indeed, if retailers try to identify bad faith actors with stricter returns policies for example, the loyal customers will also suffer and likely go elsewhere. And if they get it wrong, there can be really bad consequences, especially in the era of social media and online reviews.
Rather than shy away from policy abuse, merchants can fight proverbial fire with fire.
AI-powered fraud prevention systems can analyse billions of data points and detect suspicious behaviour patterns in real time. This allows retailers to approve ‘real’ customers’ purchases and decline those suspected of being fraudulent, in real time. Even for first time customers, and previously unseen tactics, these models can learn what ‘normal’ customer behaviour is and identify anomalies without relying on fixed rules.
Armed with the right AI technology, retailers have the power to be more targeted with their risk management to protect loyal customers and cast out fraudsters. It’s an ongoing fraud war, and the retail sector must be smarter, faster, and more adaptable than the threats they face.
If AI is powering ecommerce’s future, it must also power fraud prevention.
