Sibos 2019: Banks must improve data handling to root out financial crime
Banks need to improve the handling of their data before they attempt to apply artificial intelligence (AI) or machine learning systems to root out money laundering and financial crime, according to a panel on the second day of Sibos 2019 in London.
“When you go looking for financial crime it is a very fringe phenomenon in a very wide market,” says Marc Fungard, global head of research and analytics at HSBC. False positive rates are not at a place where they could or should be, he adds, and a better understanding of the context around data is needed.
Google says that the false positive rate with the systems some banks use verges on 100%
For Salla Franzen, chief data scientist at SEB, banks need to stop modelling outliers and start modelling behaviour. She says that understanding the networks of clients using analytics tools is crucial help see how things are connected.
Stuart Breslow, managing director and head of financial services at Google, says that the false positive rate with the systems some banks use verges on 100%.
“Walking into work and knowing that you’re clearing all of these alerts just doesn’t motivate you to get out of bed in the morning. Very little financial crime is being detected. Smart criminals know exactly what the systems are looking for, while dumb ones tend to do things that lead them to be captured.”
When financial institutions are looking to improve their processes, says Breslow, “the first thing is data, the second is data and the third is data.” Most large financial services providers are a consequence of mergers and acquisitions and a lack of investment in tech and data, he adds, so the biggest problem is solving for the data.
David Hardoon, special advisor for artificial intelligence at the Monetary Authority of Singapore (MAS), believes that if it were just about data and technology the industry has “enough people out there to solve it”. It’s not a matter of taking the existing review processing and placing an AI system on top, he argues, but about trying to understand the underlying requirements of the industry.
Breslow adds that “this is the moment of reinvention” for the infrastructure of the industry. “It’s a moment of opportunity to clean up a bunch of legacy systems and create a common data taxonomy and an understanding of what data one needs for this purpose.
“This is the data all these institutions need to run their business effectively. There’s growing acceptance and there is an opportunity for the reinvention of the regulatory environment.”
