Stopping Scams with Artificial Intelligence and Machine Learning

Artificial intelligence and machine learning models can help banks stop scams, prevent losses to fraudsters and improve customer experience

Fraudsters are opportunistic, and lately they are (unfortunately) finding a lot of success by targeting consumers directly. Using techniques like social engineering or impersonating a trusted source, these criminals are scamming unsuspecting folks out of billions of dollars globally, using real-time payments channels to quickly make off with their ill-gotten gains.

By convincing individuals to send money through apps like Venmo, CashApp, Zelle and others, or adding a payee for traditional ACH transactions, fraudsters are perpetuating what is known in the financial industry as Authorized Push Payment (APP) fraud (in the UK) or authorized user fraud (in the US). Consumers are not as worried about APP fraud and scams as they are for other types of fraud, but awareness is growing.

To help prevent this kind of fraud before it impacts consumers, banks and other financial institutions (FIs) can use advanced technology like artificial intelligence and machine learning (AI/ML) as well as sophisticated capabilities like advanced decisioning and proactive customer communications. Let’s explore how these advancements can reduce losses while protecting customers.

Authorized Payment Fraud – A Global Concern

APP fraud has been on the rise around the world. In the UK, it accounted for £249 million in consumer losses for the first half of 2022. And my colleague CK Leo notes in his recent post about payments fraud in Asia that “APP fraud is becoming a far bigger problem in India and across Asia as we see a boom in the use of real-time payments.”

During a FICO World 22 session that I moderated, we explored scams and other global fraud trends. One notable con was the “fake moto boy” scam, where a fraudster impersonates a representative from a bank and is able to convince victims to hand over sensitive details about their accounts – and sometimes physical items like cell phones and laptops.

We are not immune to scams in the US, and their prevalence and impact are getting noticed even in the halls of Congress. Another colleague of mine, TJ Horan, recently wrote a post about the growth of scams and the potential for regulation and noted that:

Data from the FTC shows that to date in 2022, consumers report losing $703 million from bank transfers and payments, $90.6 million from debit card payments, and $82.2 million from payments apps and services. 

Given the global growth, scale and incredible variety of scams happening, what can consumers, as well as banks, do to prevent and detect this kind of fraud?

Going on the Offensive in the War Against Scams

For consumers, it starts with awareness. According to recent consumer research that FICO conducted, only 5% of US consumers are most concerned about being tricked into sending a payment to a fraudster. That low level of concern seems at odds with the actual losses consumers are facing.

Understanding how scammers use social engineering and urgency to coerce immediate payments is a critical first line of defense. Consumers may also benefit from deeper understanding of the limitations in protection between P2P payments channels and other payments products like cards.

Banks can go on the offensive too. Working to educate their customers is a good start. But there are other significant actions that, when implemented as a part of a layered approach to fraud prevention, can make a big difference in protecting consumers and lowering false positives.

Since the actual consumer is the one initiating and carrying out the transaction, typical fraud detection tactics and scores simply don’t work. FIs need a different perspective to identify when APP fraud is happening.  At FICO, we’ve developed a sophisticated AI/ML model to help identify scams. Available out-of-the-box, the AI-powered scam detection score leverages behavioral profiles to identify "out-of-pattern" behavior like a new recipient or larger-than-usual payment amount to identify on average 24x the number of scams, even on a favorite device (phone, table, computer, etc.), compared to a traditional fraud score. This model just won the Machine Learning Award at the Credit & Collections Technology Awards last month.

FICO Award for Machine Learning

Other Considerations

Beyond scam-specific scores, fraud fighters can keep a couple of other considerations in mind when developing next-generation scams prevention strategies. First, think about using separate rules and strategies to address the different exposure types.

What works for ATO won’t work for an authorized push payment fraud scenario, so building separate workflows for each can help tailor your institutional approach. 

Second, consider a different customer communications strategy. Proactive communications can be a powerful intervention tool when a transaction is suspected to be fraudulent. Enabling the customer to cancel the transaction can also help improve overall customer experience, by demonstrating a commitment to putting the customer at the center of the relationship.

How FICO Helps Stop Scams and Protect Customers


Follow me on LinkedIn for more insights about FICO’s efforts to stop scams and protect customers. 

chevron_leftBlog Home

Related posts

Take the next step

Connect with FICO for answers to all your product and solution questions. Interested in becoming a business partner? Contact us to learn more. We look forward to hearing from you.