Advanced AI Helps Banks Double APP Scam Detection

  • Artificial Intelligence
  • 28.11.2022 12:00 pm

Artificial Intelligence (AI) could help banks to prevent over 100,000 Authorised Push Payment (APP) scams each year, according to a trial piloted by two major banks and LexisNexis® Risk Solutions.

The trial saw detection of in-progress APP scams double from 3-in 10 to 6-in-10 cases. Preventing 60% of APP scams before they happen could save the banking sector over £600 million in fraud losses, based on UK Finance figures for 2020 and 2021.

Recent research by LexisNexis Risk Solutions shows a third (33%) of UK financial services organisations expect APP fraud will grow over the next 12 months as the cost-of-living crisis leaves more customers vulnerable to scams.

To increase APP scam detection, analytics experts at LexisNexis Risk Solutions brought together previously unrelated intelligence - Live Call data and Global Beneficiary Information - alongside globally-sourced device intelligence. When used in combination, these data sources can provide strong indication that a scam is underway. 

Live Call data lets banks know their customer is speaking to someone on the phone whilst making a mobile banking payment, by using signals from the user’s device. Analysis shows that one in four (25%) APP scams involve a live call, making this intelligence extremely useful in fraud prevention strategies adopted by Banks.

Global Beneficiary Information can help detect whether a customer is trying to pay into a fraudster’s bank account by assessing any suspicious activity associated with the destination account.

The Live Call data and Global Beneficiary Information was then combined with unique device intelligence from the LexisNexis® Digital Identity Network® to determine whether a customer might be at a higher risk of being scammed, for example if the device has recently been used in online marketplaces or dating sites.

Using previous examples of confirmed fraud as a benchmark, AI was then able to combine these data insights to more confidently spot potential fraudulent behaviour in live customer transactions.

Director of Analytics at LexisNexis Risk Solutions, Oskar Wismierski, says: “most anti-fraud technology identifies anomalies associated with account access, which does a good job of identifying unauthorised push payment fraud. To tackle the surge in APP fraud however, we needed to look outside the banking sector for other contributory insights that can help assess the potential risk of a money transfer being made, such as whether the bank account being paid into has only just been set up, or has recently seen high-risk payments, and whether the customer is on an active phone call while making the payment, or has been phoned by an unknown caller recently – two factors linked to common types of APP.

“Combining these unique insights with device intelligence from the Digital Identity Network® to indicate other relevant online activity, gives us a much broader range of information to act upon. AI’s ability to quickly and confidently model the information against known fraud cases from the past, means that suspicious behaviour can be flagged immediately, and banks can intervene before a customer even authorises the payment to the fraudster.

“Through these trials we’ve shown that potential scam activity can be predicted with a high rate of confidence and stopped before the money ever leaves the victim’s account. These actionable insights could give UK banks a significant advantage in the ongoing fight against UK fraud that costs victims hundreds of millions of pounds a year.”

Related News