PNB turns to AI to curtail loan frauds2 min read . Updated: 29 Oct 2020, 08:30 AM IST
- Bank to deploy a system that will crawl the web for info on borrowers
- Some of the alerts sought by PNB are AI, web crawling, and optical character recognition
Lenders such as Punjab National Bank (PNB) are looking to deploy an artificial intelligence (AI)-enabled early warning system that will crawl the web for information on borrowers’ activities, including news articles and social media interactions.
Having burnt its fingers in the Nirav Modi scam and seen quite a few wilful defaults, PNB has decided to outsource its early warning signal (EWS) system. So far, the bank has been using an in-house system but wants it revamped.
Some of the capabilities sought by the bank are AI, web crawling, and optical character recognition (OCR) for triggering the early warning signal. The bank has put out a public notice seeking bids from interested parties. According to the notice, the bank wants a solution capable of giving early warning signals on a dynamic basis, based on information collected from various internal and external sources. This, the bank said, will aid timely corrective action in such loan accounts.
Indian banks have not been very quick in detecting frauds, including the ₹11,000-crore Nirav Modi case, which went undetected for years.
Worried about rising fraud in banks and delays in these being reported, the Reserve Bank of India (RBI) in 2015 took steps to address the problem. Based on the recommendations of an internal working group, the central bank introduced the practice of red-flagging a loan account as part of an early-warning system aimed at helping banks identify suspicious accounts.
“These signals in a loan account should immediately put the bank on alert regarding a weakness or wrongdoing which may ultimately turn out to be fraudulent. A bank cannot afford to ignore such EWS but must instead use them as a trigger to launch a detailed investigation into a red-flagged account," RBI said in 2015.
Meanwhile, PNB reported bad loans of over ₹1 trillion or 14.11% of its aggregate loans as of 30 June.
Some of the alerts sought by PNB are for non-achievement of projected financials given at the time of sanction; if the financials submitted to a bank are different which were submitted to regulatory authorities; any time or cost overrun or any negative development which may hamper cash flow from the project; and verification and genuineness of receivables and debtors including overseas debtor, among others.
The central bank has also acknowledged the application of AI to the fintech industry in the form of automated data analysis, chatbots and robot-advisers. “AI is being employed to detect fraud by monitoring patterns of customer behaviour," it said in June.
Experts believe India trails some other nations in AI usage. While adoption of AI in finance and banking in India has been growing, most of the applications are still support-related and not meant for decision making, unlike China where many of the critical tasks like loan sanctions are now being handled by AI, Dilip Asbe, chief executive of National payments Corporation of India (NPCI) said earlier this month.