Slow the Spread of CARES Act Fraud with the Right Tools
Times of crisis and uncertainty create desperation, so it’s no surprise that fraud has run rampant with the introduction of the CARES Act. The $2.2 trillion economic stimulus bill was created to provide financial support for businesses and individuals and preserve jobs and small businesses during the sudden economic fallout.
However, fraudsters who are eager to collect more than their fair share are creating unique challenges that must be addressed. Every dollar obtained through fraud is a dollar not going to an individual who could rightfully use the financial assistance.
A recent webinar co-hosted by DataVisor and PwC explored some of the challenges created by the CARES Act and what financial institutions can do to slow the spread of CARES Act fraud. Here’s a look at some of the highlights:
Challenge: Eliminating Friction for Good Customers without Incurring Fraud Risk
The core purpose of the CARES Act was to put much-needed funds into the right hands to prevent economic collapse. Because time was an essential factor, the barriers to entry were relatively low and made for very little friction for customers. For valid applications, a friction-free process is a welcomed experience, but it also puts financial institutions (FIs) at risk for fraud.
Conversely, implementing too many “failsafe” measures can add friction to the process and potentially result in a higher false-positive rate. Ideally, FIs should look at all data points in real time to detect patterns and subtle correlations that can allow valid customers to pass while preventing fraud from moving forward.
Challenge: Fraud Is Evolving in Real Time
DataVisor Co-Found and CEO Yinglian Xie notes that one of the greatest challenges for banks is dealing with adversaries in real time. CARES Act fraud is rapidly evolving, and fraudsters are using technology to commit their crimes just as banks are using technology to defend themselves.
“Fraudsters are using cloud centers and proxies to hide their origins, as well as device emulators and fake accounts using different email addresses,” she explains. Because of these multiple layers, FIs must not only be concerned with the fraud they find, but also with what they don’t find.
How Unsupervised Machine Learning Helps to Slow the Spread of CARES Act Fraud
Rules-based fraud detection methods are reactive rather than proactive. For banks battling new territory in CARES Act fraud, the time-consuming process of data training, labeling, and modeling could allow multiple instances of fraud to occur before their rules-based models and tools catch up.
Because unsupervised machine learning (UML) doesn’t rely on rules-based methods and data training, it enables FIs to keep pace with the evolving landscape of CARES Act fraud, including identifying previously unknown fraudulent acts. Skipping the need for data training and modeling allows FIs to respond faster to fraud before it can create financial or reputational damage. UML powers the DataVisor platform to detect high-level patterns and clusters of fraudulent activities that may evade supervised machine learning and rules-based systems.
The result is an evolving fraud model that doesn’t rely on labeling or training and can detect holistic patterns in real time with greater accuracy and less friction for the customer.
To learn more, watch the full on-demand webinar hosted by DataVisor and PwC.
