Summary of this article
NPCI non-executive chairman Ajay Choudhary urges responsible AI for inclusive finance.
Warns of bias, data dependence, and concentration risk.
Calls for stronger governance and diversified AI infrastructure.
AI is no longer science fiction in the world of finance. From loan approvals to real-time fraud detection, AI now drives nearly all of financial decision-making. Banks and other financial institutions are increasingly using AI to assess risk, manage investments, and interact with customers, transforming the way services get delivered and consumed.
But increased dependence on machine-based operations also increases the risks of bias, which could be detrimental to greater financial inclusion, said Ajay Kumar Choudhary, independent director and non-executive chairman of National Payments Corporation of India (NPCI) at the Global FinTech Fest 2025 in Mumbai. He said AI has shifted “from the margins to the mainstream”, and it is bringing along opportunities and risks for which there is a need for careful monitoring.
Between Efficiency and Accountability
Adoption of AI by financial institutions has been rapid, but regulation has lagged behind. AI systems that are applied for assessing creditworthiness, raising suspicions of money laundering, or automating investment actions are normally trained on data, but that could be already biased socially or geographically. Unchecked, they can be amplified at scale and, in the worst case, discriminate against marginalised borrowers or overestimate financial activity, he said.
Choudhary cautioned that even as AI holds out the promise of efficiency, it risks exacerbating divisions if used irresponsibly. “The industry is no longer debating whether AI will matter; it is getting ready for how much and how quickly it will move us,” he said. The question today, he added, is whether such power is being used cautiously enough to render financial systems robust and not fragile.
Infrastructure Dependence and Concentration Risks
Another less argued but more important problem is the concentration of the AI infrastructure itself. The ecosystem is based on a concentrated cohort of chipmakers, cloud providers and foundation model creators. This kind of concentration is of economic and security interest, primarily for emerging economies that rely on imported technology and data infrastructure.
Choudhary said this is one of the most urgent strategic issues in AI. He said that about 90 per cent of leading-edge processors are manufactured by a single firm, three firms control most of the world’s cloud capacity, and no few have a grip on foundation models. All such consolidation impacts not just financial stability, but economic sovereignty and national security as well.
He said “responsible AI is not a slogan; it is the only way forward”, and national and global cooperation is needed to diversify interdependencies and enhance data governance.
Balancing Innovation With Regulation
He added that regulators across the world are now seeking ways that boost AI innovation and also temper its risks. India, the UK, and the EU have all put guidelines in place on algorithmic transparency, data governance, and use of AI in payments and credit. But global coordination on a constant basis also poses problems.
He added that AI must be built and implemented on a people-first approach. He added that the “true test of AI will be whether it makes finance more stable and inclusive, not more fragile.” His statement highlighted that technological progress should go hand in hand with ethics, particularly with systems that affect the lives of millions of users every day.
India’s Digital Infrastructure
He said India’s own experience with digital payments provides a case study in how technology can drive inclusion when combined with public infrastructure and sound governance. The Unified Payments Interface (UPI), developed by the National Payments Corporation of India (NPCI), handles more than 20 billion transactions every month. Inclusionary offers such as UPI Lite in low-bandwidth regions, UPI 123Pay on feature phones, and UPI for Her for women entrepreneurs have enhanced access to digital payments for rural consumers.
Choudhary said NPCI’s federated AI models are being piloted to detect fraud without sharing sensitive user data, while real-time detection systems are improving payment integrity. Upcoming initiatives such as Bharat Connect’s Internet and Mobile Banking Platform (IBMB), which aims to standardise merchant payments, and NPCI Tech Solutions, which will function as a hub for deep-tech experimentation are steps in this direction.
“If we implement it with vision, discipline and responsibility, it can increase access, build resilience and bring growth more inclusively,” he further said.
