How does ABN AMRO detect criminal transactions?

News article
Article tags:
  • Detecting Financial Crime

New clients we investigate, for instance, by searching public sources such as media archives and convictions, using state-of-the-art software supporting information from 100,000 public resources from 240 countries.

If our investigations reveal that a company or private individual committed fraud, we might well reject them as a new client, but we might also opt to put them in a higher risk category if we consider the risks manageable. A higher risk category implies that we will keep an ultra-sharp eye on this client and their transactions.

We monitor all transactions carried out by our current clients, both through our employees in their direct contacts with clients and by using smart software flagging all kinds of deviating transaction behaviour. Our algorithms find hundreds of such transactions a day.

If we think such a deviation is worth investigating further, we may pick up the phone and ask clients to explain. With some clients, we find there’s absolutely nothing fishy going on, but others may not have solid explanations for transactions – and we may discontinue our relationship with them if we reckon the risks are too great. If clients have no or unsatisfactory answers to our questions about an unusual transaction, we will report it to our Financial Intelligence Unit (FIU). They will investigate further and may end up reporting it to investigation services such as the police, the Dutch anti-fraud agency FIOD or the Public Prosecutor’s Office.

Between 1.5% and 2% of the unusual transactions we investigate turn out to be reportable. That said, we still fail to spot criminal transactions, as criminals like to retain both their money and their freedom, and keep coming up with new ways that our employees and/or algorithms don’t yet recognise.

To stay ahead of the game, we test our methods and algorithms every year. Meanwhile, our software is part-based on artificial intelligence and is getting ever smarter in detecting transactions that need investigating.