A team of computer scientists from King’s College London has built a tool that quickly and accurately detects money laundering and is capable of scanning 50 million transactions in under a second.

The new tool for detecting money laundering is based on algorithms that rapidly identify instances where criminals divide large sums of money into several smaller transactions among many bank accounts in a technique otherwise known as "smurfing."

According to the researchers, the algorithms operate on data culled from multiple bank accounts that are represented as nodes on a complex graph that the software is programmed to focus on — detecting the most suspicious activity among those accounts.

The researchers explain that an example of this is in the deposit of $1 million. The software can reportedly monitor where that exact sum of money is being transferred, regardless of the various combinations of related transactions that occur with that money.

Considered to be three times more effective than current detection methods, which tend to be rule-based or machine-learning (ML) based, the new approach can identify the best solution for detecting common classifications of smurfing attacks across millions of amounts of data, the researchers explained.

"Our tool is also more automated and enables a far more rapid analysis of the data than what's currently available. By allowing money laundering experts to survey vast amounts of data faster than ever before, we can empower them to pick up on actors with bad intent efficiently.

"We are now working to improve the tool further, with the aim of delivering a quicker speed than conventional approaches but with even higher accuracy."

Further, the software for the new tool, which is open source, can be used with significantly larger amounts of data than current detection methods. As such, the new tool can analyze greater amounts of data over longer periods of time, notifying banks in the event it detects suspicious activity.

The tool is detailed in the study, Heavy Nodes in a Small Neighborhood: Algorithms and Applications, which was presented at the Proceedings of the 2023 SIAM International Conference on Data Mining.