INETCO in the News: Commentary: How Emerging Technologies Can Help Solve EBT Fraud

Ali Solehdin – – August 11, 2023

Instances of electronic benefits transfer (EBT) fraud have become more prevalent in the United States as criminals become more brazen and increasingly prey on society’s most vulnerable.

Supplemental Nutrition Assistance Program (SNAP) beneficiaries are one of the largest targets for EBT fraud. Last year, SNAP distributed over $113.9 billion to nearly 22 million households across the United States. Most SNAP recipients are near or below the poverty line and rely on these benefits to make ends meet. However, outdated card technology has paved the way for a card-skimming epidemic. EBT fraud is now estimated to be costing taxpayers up to $4.7 billion annually, according to the U.S. Government Accountability Office.

To understand the scale of the issue, let’s look at one of the states feeling the effects of EBT fraud most potently—New York. According to the U.S. Department of Agriculture’s most recent SNAP state activity report, in 2020 New York issued the fourth-most SNAP benefits in the country (trailing only California, Florida, and Texas). The state distributed more than $5.1 billion to more than 2.6 million individuals across roughly 1.6 million households.

Despite not ranking in the top three that year when it comes to SNAP/EBT distributions, New York came second in terms of its Federal Share of Fraud Control Costs related to SNAP payments, at over $15.6 million annually. Just across the Hudson, New Jersey was a close third at around $14 million. Meanwhile, Texas ($4.9 million) and Florida ($2.8 million) lagged far behind despite their high distribution numbers.

Shockingly, also according to USDA data, half of the entire country’s SNAP investigations took place in New York as recently as 2016. The problem has become so bad that the state has issued an official EBT Scam Alert to inform the public.

Back in 2012, benefits fraud was estimated to be costing American taxpayers just under $367 million annually compared with the $4.7 billion we’re seeing today. This increasingly pervasive criminal activity has clearly become a national crisis. The obvious questions are, what has created this surge in fraudulent activity, and how can we address it as quickly as possible?

SNAP recipients receive their benefits on a mag-stripe payment card rather than on more modern alternatives that are embedded with EVM chips. Criminals can more easily clone these cards with card skimming devices they install on ATMs and point-of-sale systems. Once cloned, scammers use the duplicate cards to extract all the benefits associated with that card.

While several states are exploring a switch to cards with embedded chips, this crossover would likely take several years to implement, by which time criminals will inevitably have developed new schemes to overcome the upgraded technology. Meanwhile, states could be susceptible to more than $9 billion more damage in EBT fraud based on the annual figures we’re seeing today.  

In addition to card skimming, fraudsters also leverage account takeovers through social engineering or purchasing bulk account information through channels like the Dark Web after data breaches have occurred. Neither card skimming nor account takeovers are particularly sophisticated in most cases, but, as evidenced by the volume of EBT fraud we’re seeing, they are unfortunately highly effective.

While the growing epidemic of EBT fraud may seem insurmountable, solutions already exist that can greatly mitigate this criminal activity. States and government agencies can integrate real-time fraud-prevention tools underpinned by artificial intelligence and machine learning into SNAP processors. This would significantly address the issue of EBT fraud in short order.

By creating individual machine-learning models that it assigns to each EBT card, this technology can assess the risk of each transaction in real time, and with far greater accuracy than alternatives. These models can account for a variety of different data points, including card-usage locations, card-load and usage timeframes, typical purchase volume, typical spends, card usage in two distinct locations in a short timeframe, card usage from a geographically distant location, and other behavioral assessments.

Automated fraud-detection tools can also be deployed rapidly without requiring either a data scientist or deep technical expertise, and can complement other fraud-prevention solutions.

Not only could automated fraud-detection technology save taxpayers and state agencies billions of dollars, it would also free up countless resources that could be directed to other areas of need.

Perhaps most important, these tools can also go a long way toward reducing the number of SNAP recipients who are having the benefits they rely on each month siphoned out of their accounts and sent into the hands of bad actors.  

Post originally published on: