Shaping the Future of Fraud: GenAI’s Stealthy Impact on Synthetic

Generative AI (GenAI) is still in its infancy but it’s already clear that it will change the way organized fraud groups commit their crimes. GenAI’s ability to generate realistic looking documents, images, and even audio and video means it can be used to commit synthetic identity fraud that’s harder to detect than current synthetic identity fraud.

Some synthetic identity fraud is committed with completely fabricated identities. The rest is built on stolen identities of real people, mixed with fabricated data. This kind of real-person-based synthetic identity fraud may not be discovered for years—especially in cases where fraudsters steal children’s data to synthesize identities. Synthetic identity fraud is a long game, and GenAI combined with automation makes hard-to-spot fraud easier to perpetrate at scale.

By 2024, synthetic identity fraud will cost banks and businesses almost $5 billion. It’s important for businesses to understand the ways in which GenAI is changing the way payment fraud happens, so they can protect themselves and their customers.

Why GenAI Can Lead to Hard-to-Prevent Payment Fraud

Banks and fintechs are on the front lines when it comes to fighting synthetic identity fraud. Nearly half of fintech companies have had to deal with fake documents as part of identity fraud scam attempts, and GenAI can make documents that are convincing enough to convince not only algorithms but also experienced fraud analysts. After banks and fintechs approve these seemingly authentic customers, they eventually offer them a credit card or a loan—and that’s when the fraudsters spring into action, go on a spree, and then disappear.

Unless they’re large enough to offer credit to their customers, most retailers and service providers won’t have the responsibility of detecting synthetic identities when they’re opening accounts. Instead, retailers and other businesses have a different problem: Once an identity passes a bank or fintech’s new account opening checks, it can be very difficult to tell whether that customer is real or synthetic.

A new customer with a legitimate looking history of online purchases might actually be a new, legitimate customer, or that customer might be synthetic. The only indication might come when a consumer whose identity was used to synthesize an ID discovers the identity fraud. That might lead to an expensive chargeback. If companies spend marketing resources engaging with synthetic customers, that loss of marketing spend can drive up customer acquisition costs.

As with other types of transaction fraud, a spate of GenAI-enabled synthetic identity fraud could push businesses to adopt stricter fraud controls that add checkout friction and increase false declines. This could end up costing those companies more than fraud. In fact, the overall cost of false declines comes to hundreds of billions of dollars every year, a number that far outweighs actual credit card fraud. In an international consumer attitudes survey by ClearSale, 66 percent of online shoppers said convenience is a major factor in their decision to buy online rather than in stores. If a store were to decline their order, 41 percent said they would never shop on that site again—and 32 percent would complain about the experience on social media, which could make it harder to attract new customers.

Best Practices for Avoiding GenAI-Enabled Identity Fraud

Because generative AI is so capable of creating realistic synthetic identities, it can be hard to detect, especially once the identity has established a credit history. However, implementing or enhancing some key fraud prevention steps can help retailers reduce their risk of not only GenAI-enabled fraud but other kinds of transaction fraud as well.

Look closer at customer identities during order decisioning. In addition to checking the validity of the payment method and the relationship of the payment method to the buyer, it’s wise to assess the history of the buyer’s activity online over time and within the past day or two. A recent spate of high value purchases can be a flag for several types of fraud, including a synthetic identity that’s cashing in, CNP fraud, and account takeover fraud.

Fight AI with AI. Fraud prevention solutions that use AI and ML can spot anomalies within specific orders quickly. Some can also detect anomalies across batches of transactions that can indicate synthetic identity fraud or bank account takeover fraud. For example, if you have multiple orders from different customers for the same product shipping to different addresses, each one of those orders may pass fraud screening and look like a good order. Batch analysis can see if their cards all have the same bank identification number, which could indicate synthetic or ATO fraud.

Document transactions carefully. You’ll need those records in case you have to contest a chargeback later—for example if an order was placed by a synthetic identity using the name of a real consumer. In the case of a completely synthesized identity, there’s no one to file a chargeback. However, if you can show that you took the proper steps to screen the transaction for fraud, you may be able to appeal the loss of order value to the bank that issued the card used to place the order.

Monitor your fraud metrics. Keeping an eye on your fraud prevention dashboard can help you identify potential surges as they emerge so you can take steps to stop them. Analyzing your historical data may help to detect commonalities among identities linked to fraud, so you can train your ML to look for those anomalies.

Keep an eye on AI-related fraud news. GenAI is a fast evolving technology and fraud tactics are always adapting to find workarounds for security measures. Every business should have someone whose role includes tracking fraud trends in the media, and every business would do well to build a security mindset into their culture.

These steps can help improve your company’s overall position against transaction fraud, not just synthetic identity fraud. They can also help you avoid false declines and friction at checkout so you can maintain a good shopping experience for your real customers.


Original article at: