San Diego's PointPredictive, which uses big data and machine learning to predict fraud, said more than a dozen auto lenders recently met in Austin to discuss how to expand the Automotive Lending Fraud Consortium.

The group, modeled after a consortium established a decade ago by some of the leaders at PointPredictive to tackle mortgage fraud, collects auto lender data and uses it to determine patterns of risk that may point to fraudulent activity. Members meet quarterly to share information about fraud experiences.

Thirteen lenders - which represent more than a third of U.S. auto-lending originations - are members. The roundtable talk last month was sponsored by Santander Consumer USA, which specializes in auto financing.

PointPredictive said it launched the consortium in response to annual fraud losses to lenders estimated at as much as $6 billion.

“All the lenders in attendance reported that auto lending fraud is a growing concern in their organizations and that they are bolstering their defenses in response,” said Tim Grace, CEO of PointPredictive. “With the consortium, we’re enabling lenders to share their patterns of auto lending fraud through predictive application risk scores and dealer risk scores — just like other industries have been doing for years. It’s almost impossible for a lender to make a big dent in their fraud losses working in isolation.”

The auto leaders said they planned to make an effort to bolster industry collaboration to better fight fraud. The consortium is open to all U.S. auto lenders.