Our habits have an organic quality to them.
Just look at your own bank statements. You will see a rhythm of life unfold:
- 8:25 a.m. Tuesday: A takeaway coffee at your corner coffee shop.
- 12:13 p.m. Tuesday: A bowl from the burrito place downstairs from your office. Extra £2.00 for guacamole.
- 7:35 p.m. Tuesday: An order for four pizzas because there’s a football match on and your friends are watching at yours because you have the biggest couch.
- 7:36 p.m. Tuesday: Your three friends kick in their share for the pizzas.
Scale that rhythm out to tens of thousands of customers, and you can picture the bank itself as a living, breathing thing, with its own rhythm of life. Lots of Monday morning coffee orders. Lots of bill payments at the first of the month. Significantly less activity between midnight and 8 a.m.
One thing that makes financial crime so pernicious is how it disrupts those habits. Bank card scams, authorized push payment scams, check fraud — these are all direct attacks on a financial organization’s rhythm of life.
In this article, Featurespace Chief Operating Officer Dr. Karthik Tadinada demonstrates how financial crime acts like an illness. It evolves, it mutates, it spreads. To protect financial organizations and their customers, fraud prevention needs an antidote which learns from past attacks so it can fortify itself against future threats.
The antidote, and key to financial immunization is machine learning.
Imagining financial crime as a spreadable illness
A common refrain among fraud and financial crime specialists is that scams do not really change. Whether a scammer sucks someone into a game of three card Monte or gets someone to park their bitcoin in a cloud mining platform, the basic scam is the same.
It’s the details of the scam that must adapt. This is the mutation process, same as with a virus.
Most people today will sniff out the ruse if they’re invited to play three card Monte. But people who are just learning about cryptocurrencies might not yet know about the cloud-mining platform scam, in which users are asked to invest some amount of their coins, which the scammer promises they will earn back with interest.
Financial crime must constantly evolve like this. The cloud-mining scam is just the delta-plus variant of three card Monte scam.
When a scam variant proves successful, it gets deployed again and again. Social media and networked economies provide easy pathways for that contagion. The scams attack individual people, feed off them, disrupt their lives and livelihoods, then move onto the next victim.
Scale those attacks out to tens of thousands of victims — a typical month in the UK — and you have an entire financial system that’s dealing with an illness.
The move now is for financial systems to build up some kind of immunity to financial crime.
How do organizations build immunity to those threats?
An immune system is a network of biological processes that work together to protect an organism from invading pathogens.
The system can spot all kinds of pathogens — viruses, parasitic worms, bad bacteria, cancer cells, even wood splinters. The immune system is outstanding at detecting novel, out-of-character objects in the body. It can even recognize how dangerous those novel bodies are by their surface chemicals.
That information allows the immune system to categorize an object as friend, unknown, or foe. When it detects a foe, the immune system can mobilize a response to expel the invader.
Immune response only works because the system can distinguish pathogens from healthy tissue. Without that knowledge, the immune response would have to be disproportionate and severe — e.g. temporarily shutting down the organism’s functions to flush out the pathogen.
Fraud prevention has long relied on the disproportionate and severe response model. Have you ever had a credit card blocked because the bank detected suspicious activity? That’s not a sustainable way to combat today’s fraud, which operates at an unprecedented scope and scale.
To combat scams, financial systems need the same pathogen-detection ability that an immune system has. With that ability, scam detection comes down to recognizing patterns of financial transactions as usual, usual but likely benign, and potentially harmful.
How an Adaptive Behavioral system is effective against evolving financial crimes
Here’s where machine learning comes in. This is the technology that gives fraud prevention the ability to distinguish between genuine and aberrant behavior, which is what we built our Adaptive Behavioral system to do.
With Adaptive Behavioral Analytics, a bank can learn about each customer’s typical spending behaviors.
- It can learn that, yes, it’s normal for this person to buy a coffee at this coffee shop on a Tuesday morning.
- It can learn that, yes, it’s normal for this person to get a burrito bowl with guacamole.
- It can learn that, yes, it’s normal for this person to be shopping in Edinburgh on the weekends and that, yes, this person tends to spend a week or two abroad in the summer.
This is the healthy tissue in this customer’s financial life. And so when an out-of-character purchase appears, the system learns something. It is constantly learning.
Perhaps the suspicious behavior is actually criminal. In that case, a network of processes is in place to protect the payment environment with a proportionate and immediate response. Should the system detect similar behaviors in the future, or even variations of those behaviors, it will know how to respond then, too.
This is how an antidote works in a healthy immune system, and it is an effective model for fighting financial crimes.
By embracing technology that can track thousands of transactions per second, distinguish between healthy and anomalous customer behavior, and respond to financial crime in real time, financial organizations give themselves an upper hand in their fight against scammers.
Discover more on how risk professionals can use machine learning to detect and prevent financial crime here.