Security Expert: Use Artificial Intelligence to Fight Benefits Fraud

by Scott McClallen

 

Nationwide, electronic benefits transfer fraud is estimated to cost taxpayers up to $4.7 billion annually, according to the Government Accountability Office.

In 2022, the Supplemental Nutrition Assistance Program distributed over $113.7 billion to nearly 22 million households.

The federal government entrusts states to reduce fraud in safety net programs. In March, the U.S. Department of Agriculture told all 50 states to plan to fight EBT skimmer fraud, which happens when bad actors install a card reader on top of a legitimate point of sale at a retail store.

Ali Solehdin, the chief product and strategy officer of INETCO, a company that provides fraud monitoring and prevention in 35 countries and processes more than $2.5 billion transactions monthly, told The Center Square in a Zoom interview that states should use artificial intelligence to fight benefits fraud.

“Why not implement an automated fraud detection system that operates in real-time that creates these individual models of each and every card that you’re issuing, and learns in real-time to block fraud before it actually happens?” Solehdin said.

Solehdin said some states want to implement chip cards, which are expensive and would take a few years to implement if they started now.

“That’s two years of fraud that’s going to go unchecked,” Solehdin said, or up to $8 billion of possible fraud, according to the Government Accountability Office.

The Center Square has records requests pending that seek Michigan’s amount of known SNAP fraud in 2022 and part of 2023.

Solehdin said the service, of which costs vary depending on the state’s population, the number of cards issued, and more, would cost taxpayers much less than replacing cards and stolen benefits.

“It’s a nominal cost and something that states are actively looking at and are interested in and need to implement, because quite frankly, without it, the situation is quite dire and could get worse as we move forward,” Solehdin said.

He said INETCO could provide a solution executable within 60-90 days that doesn’t require expertise such as data scientists to run and detect new fraud patterns. Moreso, their solutions capture “device fingerprints”, such as device/ATM ID, IP address, which identifies its location on the internet, card number, merchant, and transaction value, which would help stop fraud.

Solehdin said that artificial intelligence self-learns.

“If new fraud patterns come up that haven’t been seen in the wild before, we want to make sure that we can actually identify them without the need for data scientists to tune the model and train the model,” Solehdin said.

Solehdin said scammers can steal money in many ways. For example, a customer can distract a retail clerk long enough to deploy a skimmer on the point of sale, a rogue employee could place a skimmer, or a scammer could take over someone’s account via a compromised password or other personal information.

Then, unsuspecting customers can have their accounts drained after using the skimmer, which can leave vulnerable families hungry.

“When it comes to SNAP and EBT, you need to be able to take action because we’re talking about the most vulnerable in society, we’re talking about taxpayer dollars, and we’re talking about months to be able to recover, if at all,” Solehdin said.

States now can refill stolen SNAP benefits, but that takes time and money. In 2015-17, the total value of annual trafficked benefits increased to an estimated $1.3 billion, the most recent data available, from $1.1 billion annually in the 2012-14 study, according to the USDA.

Five states: Illinois, Massachusetts, Louisiana, Missouri, and Oklahoma are testing a pilot program to provide SNAP benefits on a mobile device, according to Payments Journal.

– – –

Scott McClallen is a staff writer covering Michigan and Minnesota for The Center Square. A graduate of Hillsdale College, his work has appeared on Forbes.com and FEE.org.
Photo “Female Using a Phone” by Andrea Piacquadio.

 

Related posts

Comments