India’s central bank is pushing banks to treat artificial intelligence as more than an efficiency tool. At FIBAC 2026 in Mumbai, Reserve Bank of India Governor Sanjay Malhotra argued that AI could reshape credit decisions in much the same way that UPI transformed payments—by reducing friction, widening access and making financial services more responsive to real-time data.
The message is significant because it comes from the country’s top banking regulator at a moment when lenders are under pressure to grow credit without weakening underwriting standards. Rather than framing AI only as a new source of model risk, Malhotra’s remarks point toward a regulatory posture that sees responsible AI as a capability banks should actively build.
From payment transformation to credit transformation
India’s payments system offers an obvious comparison. UPI lowered the friction involved in moving money by connecting banks, customers and merchants through a common digital rail. Lending is more complicated because every credit decision requires a judgement about repayment capacity, risk and affordability, but the same principle applies: better digital infrastructure can reduce the time and information gaps that make financial services expensive or inaccessible.
Recent reporting from FIBAC says Malhotra urged banks not to remain on the sidelines as AI becomes more capable. The central opportunity is in underwriting. Traditional credit models often depend heavily on formal income records, bureau histories and collateral. That works well for established borrowers, but it can leave small businesses, first-time borrowers and people with thin credit files at a disadvantage.
AI models can potentially expand the information available to lenders by analysing consented cash-flow data, GST filings, bank transactions, utility payments and other digital signals. The goal is not simply to approve more loans. It is to identify borrowers whose underlying ability to repay may be stronger than a conventional score suggests.
India already has much of the digital plumbing
The argument is especially relevant in India because banks do not have to build the data ecosystem from scratch. The country already has digital public infrastructure that supports identity, payments and consent-based data sharing, while the RBI has been developing systems intended to make credit data easier for lenders to access and use responsibly.
At MSME Day 2026, Malhotra highlighted the Unified Lending Interface as a way to reduce friction in credit assessment. The system is designed to bring together borrower-consented information such as GST filings, bank statements, utility records and land data through a common interface. For a small business with limited formal credit history, that can give a lender a more current view of economic activity than paperwork-heavy processes allow.
AI adds another layer. Once relevant data is available in a structured and permissioned way, machine-learning systems can help detect patterns, compare risk factors and surface cases that deserve closer attention. For banks, the combination of better data rails and more advanced models could shorten decision times while preserving human review for complex or borderline cases.
The opportunity comes with a governance problem
The RBI’s support for AI adoption does not mean banks have a free hand to automate lending decisions without controls. The regulator has repeatedly warned about explainability, bias, privacy, cyber risk and overreliance on automated models.
That is particularly important in credit because a model can create harm at scale if its assumptions are poorly designed. Historical data may contain biases. Proxy variables can produce unintended discrimination. A model trained during one economic environment may behave differently when borrower behaviour or market conditions change.
The RBI’s broader work on responsible and ethical AI in finance reflects those concerns. Its approach has been to encourage innovation while insisting that regulated institutions understand the models they deploy, retain accountability for outcomes and maintain safeguards around customer data and decision-making.
For banks, that means the competitive advantage will not come from deploying the most models. It will come from building models that can be monitored, challenged, explained and integrated into a clear credit-governance framework.
Why this matters for fintechs as well as banks
The impact will extend beyond large lenders. Indian fintechs already provide digital onboarding, alternative-data underwriting, loan origination, fraud detection, account aggregation and collections technology. A stronger push by banks into AI-led credit decisioning could expand demand for those capabilities—but it will also raise the bar for vendors.
Financial institutions are likely to ask harder questions about training data, model drift, explainability, audit trails, privacy controls and human override mechanisms. Fintechs selling AI into banks will therefore need to prove not only that their systems improve conversion or reduce processing time, but that those improvements can survive regulatory scrutiny.
There is also a strategic shift underway. In the first phase of digital lending, the focus was largely on taking existing processes online. The next phase is about changing how creditworthiness itself is assessed. That could create room for more cash-flow-based lending, more granular pricing and better service for borrowers whose financial lives do not fit neatly into traditional bureau models.
What banks will need to get right
The most important implementation question is where AI sits in the decision chain. Fully automated approvals may work for low-risk, highly standardised products, while larger or more complex loans may continue to require significant human judgement. The strongest operating models are likely to combine automated analysis with clear escalation rules and human accountability.
Banks will also need to ensure that faster underwriting does not become weaker underwriting. Real-time data can improve decision quality, but only if lenders understand what the data represents, how models weight it and when a signal stops being reliable.
For customers, transparency will matter just as much as speed. Borrowers need understandable explanations when credit is denied or priced differently, particularly when decisions rely on data sources they may not associate with lending.
What to watch next
The RBI Governor’s message suggests that the debate in Indian banking is moving beyond whether AI should be used in lending. The more important questions are how quickly banks can operationalise it, which data sources regulators consider appropriate, and what governance standards will apply when automated models influence access to credit.
If banks can combine India’s digital public infrastructure with responsible AI, the result could be a lending system that is faster and more inclusive without abandoning risk discipline. That is a much harder challenge than building a payments rail—but it is also why the potential impact is so large.
Sources
- Times of India — FIBAC 2026 remarks by RBI Governor Sanjay Malhotra, August 12, 2026
- TechRadar Pro — reporting on the RBI Governor’s AI lending remarks, August 15, 2026
- Bank for International Settlements — Sanjay Malhotra remarks on the Unified Lending Interface and MSME credit, June 2026
- Bank for International Settlements — Sanjay Malhotra on regulation and supervision in the digital age, February 2026
- FICCI — FIBAC 2026 conference information
