In 2016, between $800 billion and $2 trillion in illicit cash, equivalent to 2%-5% of global GDP, was laundered in various ways, according to the United Nations. In just one series of transactions, western banks including Citi, Barclays and Deutsche Bank were conduits for $22 billion illicitly shipped out of Russia from 2011 to 2014, according to an investigation by the Organized Crime and Corruption Reporting Project and Russia's Novaya Gazeta.
How it's usually investigated: Gayle Sheppard, vice president of the Saffron AI Group, an artificial intelligence subsidiary that Intel acquired two years ago, tells Axios that standard software models attempt to attack these industrial-scale criminal operations by studying financial transactions. But ferreting through the data can take weeks or months, she said.
What AI can offer: Intel is using a form of AI known as "associative memory," which Sheppard said incorporates transactions and suggestive personality behaviors, meaning the people and organizations that carry out the money laundering.
How it works: The software rapidly looks for hidden patterns in reams of emails, texts and other bank and insurance data that can indicate criminal behavior. Then it flags any anomalous data for human investigators, who check whether in fact money laundering is going on. Sheppard said this process can be carried out "in real-time."
Bottom line: The market for such software is banks and insurance firms that worry of reputational damage and fines should they be caught allowing money laundering. In January, for instance, Deutsche Bank was fined $630 million by the United States in a $10 billion money laundering case involving Russia.
