Stop Synthetic Identity Fraud in a Highly Digitized Market by Ron Whyte

Ron Whyte

Posted on February 7, 2019
artificial intelligence detecting fraud

As payments continue to embrace digital identity, thieves continue to evolve fraud methods. In the past, fraudsters used things like your name, birthdate, social security number or email address to commit fraud. This can be easy for bad guys to commit and difficult for victims to rectify. While safeguards are improving the ongoing arms race with fraudsters, the bad guys are constantly changing their tactics. One way they are doing this is through synthetic identity fraud.

Synthetic identity fraud brings a world of pain from all angles – a fluid mix of fake credentials and phony accounts that can overwhelm traditional identity theft tools, especially at smaller financial institutions. Research has indicated synthetic identity fraud in 2017 resulted in $800 billion in losses through credit cards alone and may account for 5 percent of uncollected debt and up to 20 percent of credit losses ($6 billion in 2016). The problem is even more acute with store credit cards and auto loans. And, according to the U.S. Federal Trade Commission, synthetic identity fraud is now the fastest growing and hardest to detect form of identity theft.

Keeping Identity Under the Radar

Synthetic identity fraud is a drawn-out process that demands regular management with high yield. To create the fake identity, crime syndicates (it is rarely a lone-hacker) pull together personal data to create a seemingly routine customer with multiple bank accounts across many institutions. They spend years gaming the system by making regular routine transactions on these accounts, often practicing good personal finance while building the fake consumers’ credit scores. Then when all seems routine, the fake customer cashes-in by borrowing at the top limit and maxing out credit cards at the same time before disappearing with the money.

Synthetic identities are often based on gaining a valid social security number, typically achieved by buying unused numbers from the dark web. Another method is to use the social security number of young children (with no credit line) to build up an identity without raising any alarm since children will not apply for credit until they are at least 18 years old. The elderly, who might not use their credit often, and the homeless, are also at additional risk

Applying AI to KYC

The problem seems easy to solve – surely this is just a matter of improving “know your customer” (KYC) at onboarding. But KYC rules are only as good as the financial institution itself. Internal processes may be compromised, especially at smaller regional banks without the resources to thoroughly investigate every new customer – especially given that the fake identity has all the requisite paperwork. However, given the complexity of the problem, the solution demands a more nuanced approach to check the likelihood that any customer could be fake during onboarding, but would also include ongoing checks throughout the life of the account.

A key part of the solution is the application of artificial intelligence (AI) engines and machine learning methods to comb through the growing repository of digital data about customers to better verify identity. By employing AI analytical techniques, the KYC process could be improved by checking on the background of each customer to flag any potential anomalies and highlight those customers more likely to be fake. Collaborative AI across institutions is perhaps the technology best suited for this challenge because the amount of data that banks will have to sift through is enormous and constantly growing. Once accounts are up and running, AI can be utilized to confirm that individual transaction histories follow normal patterns while flagging those that seem suspicious. 

Stronger Together

The problem is so large that it requires an industry-wide solution to solve. Banks have shown in the past that they can work together to tackle these kinds of endemic, industry-wide challenges. Without being able to monitor a customer across multiple financial institutions severely hampers the ability to detect synthetic IDs. Banks need to opt-in to schemes that look at spending patterns across the industry and across all financial institutions – banks give data to get data, and therefore improve the process.

Central to solving synthetic fraud is for banks to know their customers better. Some community banks are requiring customers show up at a physical branch to open a bank account or to apply for credit, trading high losses from synthetic fraud for a poorer customer experience. However, while it would be nice if we could return to the days when everyone had a personal relationship with their bank managers, that may be impractical in these digital times, especially for the largest banks. The eventual solution for thwarting synthetic fraud will depend on cooperation between banks, leveraging AI engines and lots of innovation, because the bad guys are innovating too.

Ron Whyte

Senior Vice President and General Manager Results-driven and business-minded executive with a focus on emerging technologies.