By FintechNews staff
-Artificial intelligence (AI) and machine learning (ML) are helping merchants and online platforms get smarter about detecting risk. The bad guys have high tech too. Deepfakes are gaining ground, along with synthetic identities. Major payments players are fighting back with a unity of sophisticated AI, ML and a mountain of data that outsmarts cyberthieves before they score. Think of it as robot versus robot.
-As many as two-thirds of consumers are inclined to abandon a merchant if there is a single incident of fraud or data theft. The merchants are aware of the problem and are on track to invest tens of billions of dollars in fraud prevention efforts through the next five years.
– Data science and machine learning have been integral to PayPal’s sophisticated fraud detection offerings which have made PayPal the preferred choice of payment processor globally.
-PayPal’s machine learning models can help predict in advance if a user’s card will be declined for a transaction and prevent the purchase from being completed. If a decline is predicted, the company can create a custom experience for the user that will ensure a valid purchase goes through, which has resulted in improvement of auth rates between 60-240 bps for certain merchants.
-Data science models can help the company in identifying good transactions so that it can “stand in” for purchase and make sure it goes through when merchants face technical issues that interrupt transaction processing. PayPal can also take this one step further by offering VIP stand-ins: for highly engaged and loyal customers, the company is able to “stand-in” for purchase regardless of the reason for a decline. Both scenarios minimize losses for a merchant and give the consumer confidence that their payment will go through when using PayPal.
-Just like traditional financial institutions, PayPal used logistic regression for fraud detection. However, now it leverages advanced techniques like gradient boosted trees (GBTs) to improve its accuracy of ML models.
-Recently, it has started to turn to more advanced AI tech like deep learning, active learning and transfer learning with their ‘Risk AI’ journey. As a result, it has been continually achieving 10-20% more accuracy over other traditional ML approaches in real time fraud detection.