Mumbai: The Reserve Bank of India wants lenders to ensure that algorithms used to make or materially influence credit decisions can explain why a borrower has been rejected, putting accountability and explainability at the centre of its expectations for technology-driven lending.RBI deputy governor Shirish Chandra Murmu, speaking at the Global FinTech Fest in Mumbai on Friday, said digital lending should not merely make the process of extending credit faster. Technology and better data should help lenders reach borrowers who have historically remained outside the formal financial system, making credit “better” as well as faster.“The vendor with the printed QR code is now part of the formal financial system. The questions she faces next are harder ones. Will she be offered credit on terms she can understand? If an algorithm declines her 10 application, will anyone be able to tell her why? If her account is frozen by a frauddetection system, how long before she can trade again? Those questions are the real content of trusted innovation, and how we answer them will matter more than any technology we deploy,” said Murmu.The deputy governor also cautioned institutions against being comfortable with their cryptographic dependencies. He said that while quantum computing is not an immediate threat payment infrastructure has long tech cycles and migration to new standards takes years. “There is also the “harvest now, decrypt later” risk, under which encrypted information collected today becomes accessible as computing capability advance,” said Murmu.The challenge is particularly relevant for India’s millions of small merchants who have moved into the formal digital payments system. The RBI’s concern is that algorithms can produce decisions that are statistically effective without being readily understandable. Murmu said well-designed models could identify creditworthy borrowers who might otherwise remain outside formal finance, but warned that the data used by such models could contain historical biases. Past relationships may not persist, a model may be economically inappropriate despite its statistical sophistication, and complex models may be difficult to explain to a customer whose application has been declined.The RBI’s position is also that automation does not transfer responsibility from the lender to the technology. “The answer cannot be the algorithm,” Murmu said when asking who is accountable when an algorithm makes or materially influences a financial decision. Responsibility, he said, rests with the regulated institution, while boards and senior management must understand the models they deploy, their limitations and the consequences of using them.
Share your thoughts in the comments
Be respectful · TOI community guidelines
This principle extends to the wider digital-lending ecosystem. Under the RBI’s Digital Lending Directions, 2025, technology can change the channel through which credit is delivered but does not dilute the responsibility of the regulated lender. Borrowers must know who is lending to them, the cost of the loan and its terms, while technology should widen customer choice rather than steer borrowers towards unsuitable products. Even when lenders operate through partnerships, the regulated entity remains accountable for services delivered in its name.
