Digital identity management and verification startup Jumio Corp. today unveiled a new fraud-fighting technology that uses artificial intelligence-driven predictive analytics to identify fraud patterns with more sophistication and accuracy.

Called Jumio 360° Fraud Analytics, the new service has been designed to tackle the issue of interconnected fraud — that is, fraud perpetrated by fraud rings or by individuals using the same information or credentials to open new accounts on banking sites, e-commerce platforms and sharing economy sites.

Jumio argues that a single organized crime operation or fraud ring can cause hundreds of millions of dollars in damage and with AI at their disposal, fraudsters are becoming more sophisticated than ever before, leaving businesses scrambling to stay ahead. This is where Jumio 360° Fraud Analytics steps in.

Jumio 360° Fraud Analytics tackles the issue of interconnected fraud with a new approach that looks beyond simple linkages, such as flagging someone because they are connected to a known fraudster. Instead, the new service looks at billions of data points across Jumio’s network to identify patterns based on behavioral similarities and other indicators.

Using graph database technology with a layer of machine learning, Jumio 360° Fraud Analytics groups identity transactions into clusters across the network and determines the fraud risk of each cluster. According to the company, this provides a multidimensional view of each transaction and the cross-customer ecosystem as a whole.

Jumio 360° Fraud Analytics also steps in when a user goes through the ID verification process. With the new service, identity transactions will also be compared to the clusters and generate a predictive fraud score that can be used to reject the transaction if it exceeds a certain threshold.

In testing, Jumio said, this approach improves its existing, highly accurate fraud detection rate by at least 30% — without increasing the false rejection rate. Business users can audit the reasons behind the decision in the Jumio Portal or via application programming interfaces.

“Because explainability is a key requirement when using automation, we provide insights on why the identity transaction was rejected,” said Philipp Pointner, Jumio’s chief of digital identity. “We also provide a graphical representation of the high-level linkages we found, which makes it easy to see connections between people, documents, devices and more.”

Jumio 360° Fraud Analytics is currently available in early release to select Jumio customers, with a general release expected in early 2024.

Image: Jumio