NextFin News - Reserve Bank of India Governor Sanjay Malhotra has warned the country's fast-growing fintech sector to treat customer data protection as the foundation of its business, not a compliance afterthought, as artificial intelligence use spreads across lending, payments and customer service. Speaking at the Global Fintech Fest 2026 in Mumbai on September 8, Malhotra said the central bank has "promoted responsible innovation while safeguarding trust, which is the very foundation of finance," framing data stewardship as the price of admission for a sector that now processes tens of billions of transactions a month on public digital infrastructure.
The governor's remarks land at a moment when India's fintech story is both a global showcase and a mounting supervisory challenge. The Unified Payments Interface, built on rails the RBI oversees, handled a record 2,366 crore transactions worth ₹29.88 lakh crore in July 2026 alone, and accounts for 84% of India's digital payments in FY2025-26. Yet the same data-rich plumbing that made India the world's largest real-time payments market - nearly half of global real-time volume, by the International Monetary Fund's count - also feeds the AI models fintechs are rushing to deploy, and with it the risk that sensitive financial data is collected beyond necessity, retained too long, or used for purposes customers never agreed to.
The Speech: Innovation Welcomed, Trust Policed
Malhotra delivered the welcome remarks in the inaugural session of the seventh Global Fintech Fest, held September 8-11 at Mumbai's Jio World Centre and Trident BKC, under a theme that itself signals where the industry is heading: "Potential to Impact: Agentic AI | Tokenisation | Quantum - Trusted, Connected, Global Systems for Inclusive Finance." Prime Minister Narendra Modi inaugurated the festival, marking his third consecutive appearance - a signal of how central digital finance has become to India's economic agenda.
"This has been made possible by fintech, which bridges and connects our vast, diverse population to formal finance faster and more affordably," Malhotra said. "Equally significant has been fintech's role in improving customer experience and efficiency in our financial infrastructure, reducing account opening and payment settlement times, cutting transaction costs, strengthening fraud detection through AI-driven analytics, enabling real-time supervision."
The festival's scale underscored the stakes. Kris Gopalakrishnan, chair of GFF 2026, co-founder of Infosys and chairman of Axilor Ventures, said the event was expected to draw more than 100,000 delegates, over 8,000 participating institutions, more than 1,200 speakers and over 500 investors. In a room that size, the governor's message was unmistakable: the RBI will not stand in the way of experimentation, but the boundary is trust.
"We have been open to new ideas and new players. We have promoted responsible innovation while safeguarding trust, which is the very foundation of finance," Malhotra said.
The Data Problem: AI Is Hungry, and Consent Is Fragile
The data-protection warning is not new for Malhotra, which is precisely what makes it significant. A month earlier, in his inaugural address to the FIBAC 2026 conference in Mumbai, the governor laid out the mechanics of the risk in unusually direct language. "The fifth risk is data privacy and security," he told bankers. "AI systems are hungry for data, and the temptation to feed them more than what is necessary, retain them longer than what is essential, or use them for purposes beyond what the customer consented to, will be constant."
That sentence captures the transmission channel behind the GFF warning. Fintechs build AI models - for credit underwriting, fraud detection, customer support, collections - by ingesting transaction histories, account behavior, device fingerprints and identity documents. The commercial incentive is to collect more, keep it longer, and reuse it across products. The supervisory concern is that each of those steps widens the blast radius when a breach occurs or a model misbehaves. Malhotra's answer was to raise the floor: "Compliance with the Digital Personal Data Protection Act is the floor, not the ceiling, of what customers should expect from their bank."
The timing matters. On July 15, 2026, the RBI released a draft "Guidance on Regulatory Expectations for Data Governance" - a 314 KB document open for comment until August 17 - that would require 11 categories of regulated entity to stand up a dedicated Data Function headed at Chief General Manager rank, create a board-level data governance committee, and name accountable Data Owners, Data Stewards and Data Custodians. The draft draws on the Basel Committee's BCBS 239 principles on risk data aggregation, and it converts what has often been contractual assurance into auditable control: the regulated entity remains responsible for data shared with or processed by third parties, with traceability back to a single source of truth, need-to-know access limits and periodic audit of the arrangement.
For fintechs and their bank partners, the message is that governance cannot stop at the corporate firewall. Much of India's fintech activity runs through partnerships - lending service providers, co-lending stacks, account-aggregator flows, analytics vendors - and the RBI's draft makes clear that outsourcing a function does not outsource accountability for the data inside it.
The Readiness Gap: Fintechs Are Scaling Faster Than Their Controls
The urgency behind Malhotra's warning is backed by the RBI's own diagnostics. The central bank's FREE-AI Committee - constituted in December 2024 and reporting on August 13, 2025 with the Framework for Responsible and Ethical Enablement of Artificial Intelligence - surveyed banks, NBFCs, fintechs and technology companies and found adoption still early and safeguards thinner still.
Only around 21% of surveyed entities were using or developing AI systems, with adoption concentrated in large public and private sector banks and NBFCs running simpler rule-based models. Among the 127 entities that reported using AI, just 15% used interpretation tools, only 35% validated models for bias and fairness - mostly at early development stages - and a mere 10% had bias-mitigation protocols in place. Eighteen percent maintained audit logs; 21% monitored for data or model drift; just 14% conducted real-time performance monitoring. Only one-third of respondents, mainly large banks, had any form of board-level oversight for AI.
The use-case mix shows why data governance cannot be an afterthought: customer support (15.6% of AI use), credit underwriting (13.7%), sales and marketing (11.8%) and cybersecurity (10.6%). Three of those four sit directly on personal financial data, and credit underwriting is the one where a biased or opaque model can deny a borrower access to capital with little explanation.
The barrier list reads like a checklist for the data rules now arriving: high implementation costs - especially binding for smaller entities - an AI talent gap, lack of high-quality training data, insufficient compute access, and legal uncertainty. The committee noted that most smaller entities preferred quick-return use cases, wary of AI's rapid sophistication and the cost of obsolescence. That is exactly the profile of a sector where third-party AI vendors and external large language models will do the heavy lifting - and where the governor's FIBAC warning that "governance cannot just stop at our outer walls" bites hardest.
The Fraud Backdrop: Why Trust Is Not Abstract
The cost of getting data governance wrong is already visible in the fraud statistics the RBI tracks. Reported cyber fraud cases in India rose from 10.29 lakh in 2022 to 15.96 lakh in 2023, 22.68 lakh in 2024 and 28.15 lakh in 2025. Estimated losses jumped from ₹6,204 crore in 2022 to ₹7,465 crore in 2023, then surged to ₹22,845 crore in 2024 before stabilizing at ₹22,495 crore in 2025 - a pattern the central bank attributes less to fewer scams than to faster intervention, transaction freezing and improved detection systems.
The international picture is consistent. An IMF working paper published in 2026 found that cyber events in the financial sector accounted for about 10% of all cyber events across 20 industry sectors in 162 countries over the past decade, concentrated in banking and securities, and that cyber-enabled fraud has nearly tripled - while remaining underestimated because of underreporting and data gaps across jurisdictions.
For a payments ecosystem where UPI alone averages 66 crore transactions a day, the supervisory math is straightforward: even a tiny fraud rate applied to that volume produces losses that can erode the public trust the entire digital finance project depends on. Data protection, in this framing, is not a legal box - it is the operating license for scale.
Cyclical Pressure or Structural Shift: What the Warning Really Signals
Malhotra's GFF warning sits at the intersection of two forces, and separating them matters for anyone reading where regulation is headed. The cyclical leg is the fraud surge: complaint volumes have nearly tripled in three years, and every high-profile scam puts fresh political pressure on the regulator to be seen acting. That pressure is mean-reverting - if fraud losses stabilize or fall, the intensity of enforcement rhetoric can ease.
The structural leg is different, and it is the one that should shape fintech strategy. India has built a public digital infrastructure - UPI, the Account Aggregator framework, the Open Credit Enablement Network, the upcoming ONDC - that makes financial data portable and reusable by design. AI makes that data exponentially more valuable and exponentially more dangerous. The RBI's response - the FREE-AI framework, the draft data-governance guidance, the DPDP Act compliance floor - is a regime change, not a reaction to a bad quarter. Rules written now will outlast any single fraud cycle because the underlying condition - abundant, reusable, model-ready data flowing through regulated and unregulated entities alike - is permanent.
The second-order implication is where the story gets less comfortable for the industry. The first-order read is simple: protect data better. The second-order read is that data governance is becoming a capital and capability barrier to entry. A Data Function headed at Chief General Manager rank, board-level oversight, audit logs, bias validation, model-drift monitoring - these are expensive to build and staff. The FREE-AI survey already showed adoption concentrated in large banks and NBFCs; the regulatory floor now rising will favor the same large players. Small fintechs that cannot afford the governance stack will either consolidate, become distribution fronts for bigger balance sheets, or operate in the shrinking unregulated interstices. The RBI is not trying to kill small fintechs - Malhotra explicitly said the central bank is "open to new ideas and new players" - but the cost of being a regulated player is going up, and that will reshape the competitive map.
There is also a cross-border dimension the industry should not ignore. UPI is now operational in 11 countries, and the IMF has recognized India's system as the world's largest real-time payments platform. If India's data-governance and AI-accountability standards harden ahead of other jurisdictions, Indian fintechs exporting their stack - or foreign fintechs entering India - will face a compliance asymmetry. The RBI's framework could become a de facto export standard for markets that lack their own, or a friction point for global platforms used to lighter oversight.
The Counter-Thesis: Is the RBI Risking Innovation It Claims to Welcome?
The strongest case against the tightening line is not that data protection is unimportant - it is that the RBI may be imposing a governance model designed for large, legacy banks onto a sector whose advantage is speed and frugality. The FREE-AI survey itself found that smaller entities preferred quick-return use cases and were wary of the cost of obsolescence; a Chief General Manager-rank Data Function and board-level committee are organizational forms that fit a 50,000-employee bank, not a 50-person fintech. If compliance costs rise faster than revenue, the result could be precisely the opposite of the RBI's intent: less innovation, more concentration, and a shadow ecosystem that migrates activity to less-regulated entities or offshore jurisdictions.
The counter-thesis has a named anchor in the RBI's own committee findings: high implementation costs were already the top-cited barrier to AI adoption, and the talent gap was second. Layering prescriptive governance on top of those constraints risks locking in the incumbents the regulator says it wants to challenge. A lighter, outcome-based approach - certify the model's behavior, not the org chart that built it - could achieve the same consumer protection with less structural distortion.
The answer to that argument rests on the fraud trajectory. If losses had been flat, a light-touch approach would be defensible. They have not been flat: a nearly threefold rise in reported cases and a more-than-threefold jump in losses between 2022 and 2024 is the market's verdict that voluntary discipline has been insufficient. The RBI's bet is that the trust being protected is worth more than the marginal innovation being discouraged - and that a fintech which cannot govern its data should not be scaling it in the first place.
The signal that would prove the tightening wrong is specific and observable: if, within two years of the data-governance guidance taking effect, reported cyber fraud cases fall materially while fintech funding and new-entity registrations continue to grow at prior rates, the regime will have shown it can protect consumers without chilling entry. If instead fraud holds steady while small-fintech formation and venture funding contract, the counter-thesis wins - and the RBI will face pressure to recalibrate toward proportionality.
What Comes Next
In the short term, the watch item is the final form of the data-governance guidance. The draft's comment window closed on August 17, 2026; the finalized version - and any proportionality concessions for smaller entities - will set the near-term compliance agenda for every fintech with a bank partnership. The FREE-AI framework's 26 recommendations, mapped across six pillars with timelines and action owners, will also move from paper to supervision.
Medium term, the pressure point is the partnership model. As the RBI makes clear that responsibility for data follows the data rather than the corporate entity, bank-fintech contracts will be rewritten to allocate audit rights, breach liability and model-validation duties. Fintechs should expect their bank partners to push governance requirements down the chain - the regulator's expectations, delivered by the banks that hold the license.
Long term, the question is whether India's data-and-AI regime becomes a competitive advantage or a constraint. If the framework produces measurably lower fraud and higher consumer confidence, it strengthens the brand of Indian fintech abroad, where trust is the scarcest commodity. If it raises costs without visible safety gains, it hands ground to lighter-touch jurisdictions.
Scenarios: the base case is a two-speed sector - large, well-governed players absorb the compliance load and gain share; smaller players niche down or partner up. The upside case is that clear rules reduce uncertainty enough to attract institutional capital that has been waiting on the sidelines for regulatory clarity. The downside case is a compliance squeeze that consolidates the sector faster than the RBI intends, with innovation migrating to the edges of the regulatory perimeter.
Malhotra's closing line at FIBAC is the cleanest summary of where the RBI stands: "The banks that will win in the AI era will not necessarily be the ones that adopt the most AI, or the fastest. They will be the ones that adopt it with the deepest understanding of what they are deploying, the clearest accountability for its outcomes, and the strongest commitment to the customer's trust that has always been, and will remain, the true capital of Indian banking." For India's fintechs, the message from Mumbai is that the same currency now buys their license to operate.
Explore more exclusive insights at nextfin.ai.
