Stablecoin Laundering Models Flagged by Duke Researchers

Researchers at Duke University have made significant strides in detecting illicit activity in stablecoins using machine-learning models. The team built a large-scale dataset of Ethereum wallet transfers, which they used to train models that flag suspicious wallets with high accuracy.

The study, published in the Blockchain Journal, found that carefully engineered behavior-based models known as tree ensembles were more effective than complex graph approaches in detecting illicit activity. These models not only raised alarms but also separated distinct types of illicit behavior, such as cybercrime and sanctioned or frozen wallets.

This distinction is crucial for compliance teams, who must decide which wallets to act on under new regulations such as the European Union's MiCA and the U.S. GENIUS Act. The researchers argue that using mathematics to block laundering can disrupt the flow of illicit funds and raise the cost of crime itself.