The rapid integration of artificial intelligence into India’s financial sector has moved far beyond theoretical discussions, becoming a core operational reality.
However, as financial institutions accelerate their reliance on algorithmic decision-making to drive efficiency, the Reserve Bank of India (RBI) is actively establishing strict boundaries to ensure market integrity and consumer protection.
Addressing industry leaders at the fourth CNBC-TV18 Banking Transformation Summit in Mumbai, RBI Deputy Governor Shirish Chandra Murmu delivered a definitive mandate: lenders must aggressively stress-test their AI systems prior to deployment and subject them to rigorous, ongoing evaluations.
This directive is not merely a bureaucratic hurdle; it is an essential defense mechanism against sophisticated cyber frauds and unseen systemic vulnerabilities.
The central bank’s regulatory posture is clear innovation cannot come at the expense of institutional safety or customer trust.
Human Accountability and Algorithmic Transparency
At the heart of the RBI’s framework for the artificial intelligence era is a set of five guiding principles, all anchored by a non-negotiable requirement for human accountability.
The deployment of complex machine learning models often obscures the chain of responsibility, creating environments where critical financial decisions are made inside a digital black box.
Murmu addressed this directly, establishing that while a machine may compute and recommend a specific decision, a designated human must ultimately own it. If an institution cannot identify the individual accountable for an automated outcome, it has not successfully deployed a tool it has recklessly delegated its core responsibilities.
To maintain this oversight, the RBI expects every bank to maintain a comprehensive, real-time inventory of all AI systems currently in use. Crucially, this includes external algorithms deeply embedded within third-party vendor products, which often escape direct internal scrutiny.
Transparency extends directly to the consumer level as well. Customers have a fundamental right to know when they are interacting with an automated system.
More importantly, when an AI-driven process results in a material impact such as a rejected loan application, a reduced credit limit, or a denied insurance claim there must be a clear, accessible escalation route to a human authority capable of reviewing and potentially overturning the algorithm’s verdict.
Managing Systemic Risk and Expanding Credit Access
A significant pitfall for modern banks is viewing artificial intelligence exclusively through the lens of cost reduction.
Implementing automated systems simply to compress operational expenses, without simultaneously widening market reach or elevating the user experience, is a failure of innovation that merely automates existing flaws.
This narrow focus can inadvertently harm economic inclusion. Citing recent credit bureau data, Murmu highlighted a troubling metric: the proportion of new businesses successfully entering the formal credit system dropped from 52% in 2022-23 to just 42% in 2025-26, even as overall outstanding commercial credit expanded by 14%.
There is a distinct risk that conservative AI models are equating a lack of historical data with high financial risk, effectively translating “we don’t know” into “we know it’s bad,” thereby denying critical capital to emerging enterprises.
Beyond individual institutional blind spots, the shared architecture of modern banking creates severe contagion risks.
As multiple banks increasingly rely on the same foundational data sets, third-party vendors, and cloud infrastructures, isolated vulnerabilities can rapidly mutate into sector-wide threats.
When risk becomes collective, the intelligence deployed to counter it must also be collective. The central bank has already initiated this collaborative defense through platforms like MuleHunter.ai and the Digital Payments Intelligence Platform.
Reinforcing this strategic direction, RBI Governor Sanjay Malhotra emphasized that banks must actively dictate their AI trajectory rather than being shaped by default.
By thoughtfully integrating artificial intelligence with India’s robust public digital infrastructure including Aadhaar, UPI, DigiLocker, and ONDC the private sector can build a resilient framework that accurately assesses risk, fairly prices capital, and fundamentally redefines customer service.
Source: Official LiveMint, "Deputy Governor Murmu Asks Banks to Stress-Test AI Systems Before Deployment"
