Researchers tested their AI tool by comparing its findings with a cryptocurrency exchange, uncovering 52 suspicious transactions that all led to the same exchange. The exchange had already flagged 14 of these accounts for potential illicit activities, such as money laundering or fraud. Despite not having access to the exchange’s customer data, the AI model matched the investigators’ conclusions.
While identifying 14 out of 52 suspicious accounts may seem low, the researchers argue that only 0.1 percent of the exchange’s accounts are typically flagged for money laundering. They believe their automated tool significantly increased the efficiency of detecting potential illicit activities.
Elliptic, a company that specializes in blockchain analysis, has been using the AI model privately. The researchers also found connections to a Russian dark-web market, a cryptocurrency mixer, and a Ponzi scheme by analyzing transaction chains identified by the model.
The release of Elliptic’s training data on Kaggle, a machine learning community site, has the potential to improve anti-money-laundering efforts across the industry. While the current AI tool may not completely revolutionize the field, it serves as a promising proof of concept for future developments.
Overall, the researchers emphasize the importance of collaboration and open-source practices in advancing the fight against money laundering in the cryptocurrency space.
