Is someone else borrowing under your name?

Identity thefts are becoming increasingly common and without the use of technologies such as machine learning and AI, it can turn into another pandemic.
  • Updated On Aug 5, 2021 at 09:22 AM IST
Read by: 100 Industry Professionals
Reader Image Read by 100 Industry Professionals

By Parikshit Chitalkar

With every new technological advancement comes its own set of challenges. As financial transactions go digital, so has financial fraud. In fact the scale at which financial frauds are happening today is akin to an entire industry altogether.
Let’s put things into perspective. According to RBI financial institutions in India reported frauds worth Rs 1.38 trillion in 2020-21. And this doesn’t even take into account frauds that are less than Rs 1 lakh in value. OR fraud that has gone undetected, modern day Synthetic fraud is so advanced that it is near impossible to detect via manual processes

Advt
It is not just the scale that is going up but even the complexity of these frauds. This is happening despite financial institutions deploying hundreds and in cases thousands of rules to detect fraud. The problem is, these rules become obsolete within days and it takes months for financial institutions to rewrite the rules.

However, enterprises should take cue from fintech startups that are deploying machine learning algorithms to generate self-learning rules for keeping pace with new and complex frauds.

Since it is nearly impossible to guess a fraud pattern and predict the detection rules, machine learning allows data scientists to determine which transactions are most likely to be fraudulent. The automated fraud pattern discovery allows companies to detect frauds even when transaction volumes are in millions.

Machine learning achieves one more challenge apart from fraud detection—false positives. False positives can often overwhelm the systems and even force lenders to deny credit to genuine customers, leading to poor customer experience and loss of business. With machine learning false positives can nearly be eliminated, making the fraud detection engine more reliable and act as an agent of improved customer experience.

Types of fraud

New types of frauds are rapidly emerging in the lending space while the existing ones are getting increasingly sophisticated. Right from fake IDs to identity theft and fraudulent credit applications are some of the most common type of frauds lurking in the wild.
When a fraudster steals your identity to take credit in your name and address, it may get hard to convince the authorities that it wasn’t in fact you who took the loan or convince the credit bureau why there’s a massive default under your name. Chances are you’ll not even get to know there are loans running under your name until one day you go to apply for a loan and it gets rejected because someone else has ruined your credit rating.

Advt
However, by using machine learning and AI, identity theft can be mitigated. By Machine learning models can use patterns identification to significantly improve the accuracy of fraud detection. Even after a credit is issued, lenders can continue to track spending pattern to detect fraud and ascertain any possible case of identity theft. Similar AI models can be combined with image processing to identify forged documents that can prevent anyone from borrowing under a fake name or identity.

Fintechs leading the chart

The use of AI and ML for fraud detection has already picked up massive pace in the digital lending space in India. Fintech companies are deploying these machine learning models to analyze large streams of data to make decisions and detect fraud.
Fintech companies are using data scientists in the field of credit underwriting to ascertain the risk associated with lending to a customer using the least amount of customer input possible, to try and deliver a decision to the customer instantly. This not only cuts risk for the lender but also allows them to deliver a great customer experience at a low operational cost.

Companies are also deploying AI and machine learning and pair them with computer vision models to track identify fraud, which has been a growing concern. Companies are also working on deep learning algorithms to track fraudulent transactions,creating a much-needed blanket of protection to both the lenders as well as customers. Some fintechs in India are now starting to use cutting edge computer vision models to detect identity fraud using image analysis, liveliness detection & real time comparisons to known fraud data sets.

The author is Co-Founder, StashFin


  • Published On Aug 5, 2021 at 09:21 AM IST
Be the first one to comment.
Comment Now

Join the community of 2M+ industry professionals

Subscribe to our newsletter to get latest insights & analysis.

Download ETCISO App

  • Get Realtime updates
  • Save your favourite articles
Scan to download App