Hunting Down Money Launderers with AI
Photo Source: Pixabay
The diversification of banking online and the rise of cryptocurrency have given criminals various new tools and techniques to launder money locally and around the world. A new technique called isolation forest uses artificial intelligence to help banks and law enforcement track down money launderers. Money laundering refers to the process of converting "dirty" money obtained through criminal activity such as drug sales, prostitution, or arms dealing into "clean" cash by passing it through a complex series of legitimate businesses and financial institutions. The process makes it very difficult to trace the illegal origins of the money, allowing the criminal to use it as he pleases. According to the Anti-Money Laundering Forum, the term money laundering originated with the notorious gangster Al Capone, who purchased numerous laundromats across Chicago, where he comingled the legitimate money from these businesses with the proceeds from his sale of illegal alcohol and other activities. (anti-moneylaundering.org)
The National Money Laundering Threat Assessment, developed jointly by several US government branches such as the FBI and Department of Homeland Security, details various money laundering techniques from check-cashing services and stored value cards to bulk cash smuggling and shell companies. (justice.gov/dea) Add to that the emergence of cryptocurrencies such as Bitcoin that allow nearly anonymous financial transactions without a bank in between, and now criminals have an expanding variety of options open to them for money laundering.
Money laundering remains a considerable problem. The United Nations Office on Drugs and Crime estimates that between $800 billion to $2 trillion or 2-5% of the global GDP gets laundered each year. (undoc.org) Such massive amounts of money laundering demonstrates the difficulty in combatting it. As mentioned above, there are many money laundering schemes from cash smuggling to asset conversion, such as buying diamonds in one country and selling them in another, thus breaking the connection to the illegal money. In financial markets such as a stock exchange, criminals launder money through a technique called 'wash trading.' In wash trading, one person or a group will simultaneously buy and sell the same stock from different accounts. The buy and sell activity drives up market volume and price while effectively anonymizing the illegal money. Wash trading became illegal in the United States in 1936 under the Commodity Exchange Act.
Detecting wash trading among millions of trades on exchanges such as the New York Stock Exchange poses a difficult challenge for banks and law enforcement. For the most part, these trades look like any other trade, but wash trading has subtle elements that set it apart from typical trading. For example, wash trades often buy and sell almost identical amounts of stock very close in time. The close timing helps to guarantee that little money gets lost in the transaction due to market fluctuations. (visallo.com/blog) Detecting such deviant trades when distributed across multiple investors challenges investigators. However, with the aid of a type of artificial intelligence called isolation forest, investigators can now better identify wash trades among the millions of legitimate transactions.
Isolation forest refers to an irregularity or 'anomaly' detection program which a computer uses to look over large amounts of data such as stock transactions to discover patterns that separate the illegitimate trades from the many legitimate ones. The name isolation forest refers to the system the computer uses to look for patterns called a decision tree. Decision trees represent a way to divide information into different patterns of legal and illegal stock purchasing and sales. For example, certain stock traders almost always post a trade before noon or may never offer to sell at market price. Such details create patterns that isolation forest can find that represent legitimate traders and detects the wash traders' subtly different patterns. A human analyst cannot possibly look over the millions of stock trade records that a computer can, but isolation forest finds unique trees that define wash traders, making isolation forest an excellent tool for banks and law enforcement.
Criminals face a significant problem when they try to use money gained illegally. So-called "dirty money" raises flags with banks and retailers for mortgages, luxury purchases, or other large transactions without a paper trail that proves the money was earned legally. To get the cash cleaned, criminals have worked out many money laundering schemes from padding the books of legitimate businesses like restaurants, shell companies, and stock trading schemes such as wash trading. Wash trading involves the near-simultaneous buying and selling of the same stock by the same person or group with multiple, unrelated accounts, effectively obscuring the origin of illegal money. Although illegal since 1936, wash trading remains challenging to detect because the transactions look almost identical to legitimate ones. However, an AI algorithm called isolation forest can find subtle patterns to distinguish wash trading from the rest of the millions of legitimate trades. Isolation forest identifies subtle differences in wash trader behavior on which banks and law enforcement can then act. As money laundering options grow, artificial intelligence has emerged as a new countermeasure against this criminal activity.
Dr. Smith’s career in scientific and information research spans the areas of bioinformatics, artificial intelligence, toxicology, and chemistry. He has published a number of peer-reviewed scientific papers. He has worked over the past seventeen years developing advanced analytics, machine learning, and knowledge management tools to enable research and support high level decision making. Tim completed his Ph.D. in Toxicology at Cornell University and a Bachelor of Science in chemistry from the University of Washington.
You can buy his book on Amazon in paperback and in kindle format here.