Banking: Fighting fraudulent account opening using Machine Learning

Published by FirstAlign

A banks’ primary aim is to provide a convenient and secure banking experience to all its customers. The first step in acquiring customers is to offer frictionless and easy account opening options. Today, this is done at the click of a button—an attraction to both customers and fraudsters alike.

Banks are continuously innovating with the latest in technology to improve their risk management capabilities. In this article, we discuss how Machine Learning( ML) can detect account opening frauds.

Account opening fraud

Account opening fraud is when fraudsters use stolen identities to open new bank accounts.

There are two types of account opening frauds.

  1. Identity Fraud: Identify fraud is when a fraudster opens a bank account using the personal information of a person. 
  2. Synthetic Fraud: Synthetic fraud is when a fake identity is created using valid information. For example, a real social security number can have a phony name attached to it.

KPMG Global Banking Fraud Survey indicates that over half of respondents recover less than 25 percent of fraud losses, making fraud prevention vital.

Photo by Bermix Studio on Unsplash

Detecting account opening fraud is not easy.  Fraudsters obtain valid customer information from the dark web just for a few dollars. In synthetic fraud, fraudsters steal information from multiple persons to make a new identity. In such cases, it becomes even more challenging for banks to find the legitimacy of the application. Customers may not know that their information is stolen, and since details in application are individually valid, banks may find it challenging to uncover fraudulent activity.

The traditional approach depends on rules, but rules alone are no longer sufficient. Rules can be too narrow or too broad, and fraudsters can easily find conditions to breach them. Banks’ existing security systems work well when banks already have the necessary information about their customers. In new account openings, banks do not yet have the information.

Fighting account opening fraud using ML.

Rules alone cannot provide a complete solution; this is where Machine Learning comes in handy. Machine learning, along with knowledge-based authentication (KBA) methods like PINs and passwords, provide better fraud detection mechanisms. By combining Machine Learning with rules, banks can keep pace with fraudsters and identify fraud signals early. There can be many fraud signals, and Machine Learning help connect the dots.

For example, Feedzai, a data science company that detects frauds for financial institutions, has found high battery power devices correlated with high fraud. When a device name is unknown, the likelihood of fraud is  78%. Interesting insights! Isn’t it? These exciting insights provided by a machine are decisions in the making.

Humans can make better decisions when they are provided with ML analytics. ML offers speed and convenience. Users go through security checks without being alerted. It reduces the need to provide extra information, thereby decreasing the time spent on opening an account.

Account opening is the starting point of a relationship with customers. Manual verification is costly, time-consuming, and unreliable. Feedzai’s machine learning model has benefited a US Bank with a 75% increase in application approvals.

Conclusion

Given the complexity of account opening fraud, banks no longer depend only on rule-based detection methods. Machine learning with KBA provides a more holistic risk management system. It is faster, reliable, and costs less than manual processes. Advancements in Machine Learning provide greater possibilities, and banks continue to innovate to stay ahead in the fraud game. 

References

Published by FirstAlign

Click here to connect with us

Spread the word

Related posts