Fintech group MyBucks has invested in technology that is specifically aimed at exploiting big data to mitigate fraud and identity theft.
With its own in-house Artificial Intelligence (AI) team, Dave van Niekerk, CEO of MyBucks, explains that by using classification or grouping algorithms within AI, they can identify fraudulent behaviour.
As an increasing number of consumers take to transacting online, so grows the opportunity for tech-savvy fraudsters to use ever evolving ways of stealing identities to financially exploit unsuspecting consumers.
Local South African statistics indicate that an identity is stolen as often as every 29 seconds.
Our fraud detection engine, has for the last three months, already prevented more than 1500 fraudulent loan applications and continues to inhibit the registration of stolen identities every day. Because our business model is digital, we do not have face time with the client, we can remove emotional bias, and it forces us to analyse every bit of data we can obtain.
Since transactions are conducted solely online, we don’t see the client, and thus don’t know who we’re interacting with. We, therefore, need an advanced model that through capturing a lot of data can detect patterns almost like a finger print.
In most instances, fraudulent activity involves fraudsters creating accounts with stolen identities where the accounts remain dormant until they find a loophole and then activate accounts to exploit that loophole.
Our fraud detection engine uses an algorithm to analyse the data that is captured when clients interact with our channels. It then connects clients through shared online behaviour patterns such as bank accounts, IP addresses and typical email addresses for instance.
If the data patterns deviate from the norm it then raises a flag as to possible fraudulent activity. For example, the algorithm will pick up the time spent on completing a registration something which fraudsters typically do very quickly and often use keyboard shortcuts like CTRL-V.
However, when it comes to blacklisting clients, while the engine will bring up an alert when a breach is detected, the breach does go to a credit officer for further investigation and, if need be, certain cases are taken further to the police to investigate.
Over time, clients that behave similarly to fraudsters will be grouped with known fraudsters.
We are currently in the process of capturing as much information as possible.
We want to be able to take as much information we can into account when categorising clients. In its simplest form, the engine’s functionality is about detecting similarities, which is much harder than detecting equality.
Going forward, MyBucks are in the process of developing facial recognition algorithms that will be incorporated into the fraud prevention system.
Most European markets currently require copies of photo identity documents, and as such we are in the process of developing facial recognition algorithms to address this requirement.
A longer-term goal is to add biometric verification using facial recognition to our mobile applications. This can be very useful in African countries where national ID systems are non-existent or poorly functional.
At the end of the day it’s about leveraging technology to enable us to provide faster and more efficient financial services to clients.
However, at the same time, we want to ensure we mitigate all possible risk when it comes to potential fraud, and by using the benefits of big data analytics to our advantage, we are confident we will continue to make ongoing inroads into minimising and eventually eradicating it all together.