Kount Integrates AI in Fraud Prevention Platform 

Dec. 10, 2012

Antifraud software provider Kount opened the hood of its proprietary technology platform last week to give the industry a peek at part of its engine. The Boise, Idaho-based company revealed some of the workings of the artificial intelligence technology that enables its platform to learn from historical outcomes and integrate that information in fraud detection in real time.

Kount said its AI includes supervised machine learning—a specialized branch of statistical learning theory that “combines historical training data with predictive functions that process real-time threat data.” This technology has enabled Kount to “crack the nut” on applying AI to fraud detection, according to Steve Rouse, COO of Kount.

“Some of the huge challenges with using AI to detect fraud relate to the computational complexity of doing it in real time,” Rouse said. “Kount has solved these math problems, and reduced the overhead required to effectively deploy AI against evolving threats in real time. Neural networks and Bayesian models are susceptible to noise, which is partly responsible for the high rate of false positives these models can generate. Kount’s AI approach is more accurate because we deploy active filters to remove that noise and bias.”