--- title: AI Could Do for Indian Lending What UPI Did for Payments, Reserve Bank Argues description: India's central bank governor says AI reading GST filings, cash flows and utility bills could extend credit to millions with no traditional credit history. author: Darie Nani (Editor-in-Chief) updated: 2026-08-14T13:16:46.104Z canonical: https://www.sovereignmagazine.com/article/rbi-ai-lending-alternative-data-upi categories: FinTech content_type: News region: India publication: Sovereign Magazine schema_type: Article --- Artificial intelligence could change how credit reaches Indians as fundamentally as the Unified Payments Interface changed how they move money, Sanjay Malhotra, Governor of the Reserve Bank of India, told the country's bankers this week. Speaking at the FIBAC national banking summit on Tuesday, he set the UPI parallel deliberately high: India's near-universal instant digital-payments system democratised transactions for hundreds of millions of people, and the governor argued that lending is the next function ripe for that kind of expansion. The mechanism he pointed to is not faster loan processing but a wider definition of who counts as a viable borrower. ## Alternative Data Could Reach Borrowers Banks Now Turn Away The core of Malhotra's case rests on data that conventional underwriting ignores. A borrower with no loan or credit-card record is invisible to a traditional scoring model, yet the same person may file GST returns, run steady cash flows through a bank account, or pay utility bills on time. AI models trained on those signals, the governor said, let a lender judge the creditworthiness of someone the old system would simply reject. He described this as expanding the "frontier of bankability", looking past conventional financial histories to reach underserved sections of society. > "AI, well-deployed, can close existing gaps in financial inclusion faster than any preceding technological innovation." > — Sanjay Malhotra, Governor, Reserve Bank of India In his words, "AI models on alternative data such [as] cash-flows, GST filings, utility payment bills etc could help banks extend credit to customers who may not have sufficient traditional credit history." That population, the so-called new-to-credit segment, is where the governor sees the largest gap between demand for credit and access to it. ## India's Digital Rails Give Private Lenders a Foundation to Build On Malhotra's argument depends on infrastructure India has already built. He pointed to the country's digital public infrastructure, including Aadhaar, UPI, DigiLocker, the account aggregator framework and the unified lending interface, as a foundation on which private-sector lenders can layer AI applications. The claim is that these systems already move identity, consent and financial data at scale, so the marginal cost of extending credit to the last mile falls once AI sits on top of them. The result, in the governor's account, is financial services that are faster, more granular and more accessible. The lending frontier is not the only use he cited. Relationship managers equipped with AI could serve more customers with relevant product recommendations and risk alerts, and the technology could support grievance redressal and financial guidance. He drew a sharper contrast on fraud: rules-based engines fall behind adaptive fraudsters, while AI and machine-learning models can keep learning from emerging patterns and flag anomalies in real time. ## Banks Must Own Every Lending Decision The governor's caution was aimed less at the technology than at how banks adopt it. He urged them to treat AI as "a deliberate, board-driven strategy, backed by sustained investment, and most importantly, a strong intent, rather than a series of disconnected projects", and warned that lenders "cannot afford to sit on the sidelines" as adoption accelerates. He framed AI as a capability to be responsibly harnessed rather than merely a risk to be contained. He was direct about the failure modes. Careless deployment, he said, could create new forms of exclusion and instability, and banks must confront algorithmic bias, concentration, dependence on third-party vendors, data privacy and cyber threats. On who answers for a lending decision, he left no ambiguity: "No matter how sophisticated the model may be, for a bank's decision, the ultimate responsibility has to lie with the bank, and not with the vendor or the algorithm." ## FAQ **Q: What does "new-to-credit" mean?** It refers to borrowers who lack a sufficient traditional credit history, meaning no meaningful record of past loans or credit-card use for a lender to score. Because conventional underwriting leans on that history, new-to-credit customers are often turned away regardless of their actual ability to repay, which is the gap the governor wants alternative-data models to close. **Q: What is alternative-data lending?** It is assessing a borrower using signals outside the standard credit file, such as GST filings, bank cash flows and utility payment records. Malhotra's point is that AI can read these patterns to gauge creditworthiness for people the traditional system cannot see, extending the "frontier of bankability" to underserved sections of society.