India’s central bank wants AI to approve loans that humans would reject
The governor of the Reserve Bank of India, Sanjay Malhotra, has called on the country's banks to use artificial intelligence to approve loans that human assessors would typically turn down, arguing this could significantly expand financial inclusion. Speaking at the FIBAC conference in Mumbai, he said AI models trained on alternative data could assess creditworthiness for first-time borrowers, gig workers and small businesses that lack formal financial records, extending access to banking well beyond what manual underwriting allows, and at a fraction of the cost.
Malhotra suggested AI could also help identify borrowers at risk of default early enough for banks to offer guidance rather than pursue recovery, and could support voice interfaces in India's many local languages to reach populations with lower literacy rates. He also flagged risks, notably the "black box problem" of opaque AI decision-making, which makes it hard for regulators, auditors and borrowers to understand why credit is refused, along with concerns about algorithmic bias and over-reliance on a small number of AI technology providers.
In short
- RBI governor urges banks to use AI for riskier loan approvals
- AI could extend credit to unbanked, informal workers, small businesses
- Governor warns of AI "black box" and bias risks in lending
Steelman: both sides, in good faith
The strongest fair case each way — we don't pick a winner.
The case for
AI loan approval could dramatically expand financial inclusion by assessing creditworthiness for first-time borrowers, gig workers and small businesses lacking formal records—populations currently excluded from traditional banking. Using alternative data and operating at scale, AI systems can make decisions far more cheaply than manual underwriting whilst reaching lower-literacy populations through local-language voice interfaces. The technology can also identify early default warning signs, enabling preventive support rather than harsh recovery measures, thereby improving both access and outcomes for underserved communities.
The case against
Deploying AI to approve loans raises serious fairness and accountability concerns that merit caution, particularly for vulnerable populations. The 'black box problem' means borrowers cannot understand why they have been rejected or challenge decisions, whilst regulators struggle to audit compliance with lending standards, creating accountability gaps. Algorithmic bias is well-documented and can perpetuate or amplify existing discrimination, whilst concentrating lending decisions among a small number of AI providers creates systemic risk. These governance failures are especially troubling when loan approval is the gateway to economic participation.
Explained simply
Banks usually decide whether to lend by checking a person’s income, credit history and other financial records. People borrowing for the first time, gig workers and small businesses may have little of this paperwork, making them harder to assess through standard methods.
The Reserve Bank of India is the country’s central bank, which oversees banks and financial rules. Its governor, Sanjay Malhotra, has argued that AI could use other kinds of data to assess borrowers, while also warning that opaque decisions and bias can make it difficult to understand or challenge a refusal.
The issue matters because wider access to loans could help people and businesses with limited financial records, but lending decisions affect borrowers’ finances and future options. Banks and regulators therefore face questions about how to check that automated assessments are fair and explainable.