How Artificial Intelligence Can Protect Nonprofits Against Fraud

As we’re entering a very technologically advanced age, nonprofit organizations are often victim to fraud. Fraudsters are targeting nonprofits, using nonprofit donation pages as testing grounds for fraudulent card testing. Card testers build computer algorithms that automatically and rapidly attempt donation form submissions with different credit card number combinations. The goal of the fraudsters? Try to find a valid credit card number. And with over 1.8 million nonprofits in the U.S., there are plenty of opportunities for fraudsters to attack.

The evolution and progression of technology gives these fraudsters opportune tools to commit fraud, and they’re becoming progressively more sophisticated in what they do; but it also equips nonprofits with the right resources for the prevention of fraud.

Enter Artificial Intelligence (AI).

In an exclusive interview with NonProfit PRO, Alex Sukhenko, VP of Technology at Salsa Labs, and Donna Myers, Chief Operating Officer at Salsa Labs, shared how the implementation of AI protected their nonprofit clients against fraud.

Before discovering that AI could protect against fraud, it was up to individual nonprofits to ask for CVV and address validation on their donation pages in order to prevent fraud. Fearing lower donation page submission rates, many nonprofits opted not to ask for additional verification details.

“A lot of our clients were pushing back saying that the more credit card verification information that they asked from the donor, the fewer donations they were going to get,” Myers said.

That’s when Salsa Labs considered AI as a solution for their clients. Before AI, Salsa Labs had in-house capabilities that automatically detected certain patterns, which would, ultimately, block fraudsters. But as fraudsters became more advanced, the in-house tooling became “inadequate,” according to Sukhenko. This created a lack of confidence in them from their nonprofit clients, and it also created unnecessary costs for them on credit card payment processing.

What Makes AI Different

With AI, there’s a footprint pattern and a global network available. There is also a constant real-time learning algorithm that holds the capability to detect fraudulent attempts and patterns.

“We benefit from the network from already-identified fraudsters, so when someone comes to our clients’ donation pages and tries to do something—and they’ve done things like this previously on the Internet—they’ve been labeled as such that they get marked as fraud,” Sukhenko said.

In the past, Salsa Labs was able to source IP addresses. If there was a frequency of smaller transactions ($10, $20 or less), they were able to detect and shield that kind of activity. But these days, fraudsters are using various cloud services, which creates infinite IP addresses. But with AI, there’s a lot of triangulation of information on client IDs that are kept and are global, so suspicious activity can be detected.

Sukhenko noted that last year, from #GivingTuesday to Dec. 31, the company processed over 350,000 credit card transactions and over $35 million without any fraudulent breaches on client donation pages.

“Prior to that, we had seen fraud attempts happen on a monthly basis. That’s why we wanted to get proactive this giving season to make sure we protect our clients, and they didn’t have to go through anything during the busiest time of the year,” Myers said.

Looking to the future, will fraudsters continue to become more sophisticated and pick up more tactics to eventually outsmart AI? Sukhenko doesn’t believe so:

“It's kind of an arms race. Fraudsters have techniques, and we have techniques. We have improved our techniques, and, by definition, AI is sort of an evolving real-time model that will improve on itself; and especially the many customers that feed into the global model and footprint. It should become very difficult, but if it does, I'm sure we will come up with an improvement that will manage.”