RBI Pushes NBFCs to Strengthen AI Underwriting

NBFC AI underwriting — NBFC AI underwriting is moving onto the Reserve Bank of India’s supervisory agenda. Deputy Governor S C Murmu urged non-bank lenders to strengthen underwriting, stress testing and risk management while using AI and machine learning carefully.

What is verified about NBFC AI underwriting?

The regulator’s message links technology adoption to basic credit discipline. Better models cannot compensate for weak data, aggressive growth targets or poorly governed overrides.

Verified facts and evidence boundaries

Measure	Value	Status
Speaker	S C Murmu	RBI deputy governor
Priority	Underwriting	Standards must keep pace
Tools	AI and ML	Governed use encouraged
Safeguard	Stress testing	Customer protection included

What the headline does not prove

The remarks are supervisory guidance, not a finding that AI automatically improves loan quality. Lenders remain responsible for explainability, fair treatment and model risk.

News announcements mix completed events, planned milestones and attributed performance claims. This report keeps those categories separate. A release date is not delivery, a vendor benchmark is not an independent test, and a policy proposal is not an implemented rule. That distinction matters to managers making procurement, compliance or investment decisions.

How businesses should evaluate the change

Start with the operational chain: identify the data, hardware, software, people and approvals required before the headline can produce a measurable outcome. Then assign an owner and a failure mode to each stage. This exposes whether a strategy has genuine redundancy or simply several components depending on the same provider, dataset or approval path.

Next, define a baseline before adopting the new system. Teams should record current cost, error rate, completion time, utilisation and customer impact. Without that baseline, a faster demonstration can look like progress even when total workflow cost rises. Procurement should also include exit rights, data-export capability and a recovery process when the service fails.

For India, the practical questions are availability, local pricing, data residency, language support, integration labour and enforceable service commitments. A global launch does not guarantee an India release. Indian organisations should test the narrow workflow that creates value and retain human review wherever errors affect employment, safety, finance, education or customer rights.

Related Lapaas Voice reporting on Microsoft Teams helpdesk attacks and Bodhan education AI models provides adjacent operating context. Our coverage of Reolink local security AI and India aircraft leasing shows why implementation evidence matters more than a launch claim.

Source and verification note

The event and its context were checked against RBI, Economic Times, CRIF High Mark, CRISIL Ratings. Figures remain attributed to the organisation that supplied them unless an independent measurement is identified.

A decision checklist

Confirm the contractual or policy status, not just the announcement date. Verify which features are available now, which are in preview and which remain targets. Document the information that leaves the organisation, who can access it, how long it is retained and how it can be deleted or exported.

Run a limited pilot with success and stop conditions. Measure accuracy, exception volume, human review time, reliability and total cost. Compare results with the existing process rather than with a vendor demonstration. If the system touches regulated or safety-critical work, require legal, security and domain-owner approval before expanding deployment.

Finally, revisit the decision when primary evidence changes. A final filing, shipped product, incident report, audited result or regulator notice can materially alter the analysis. Updating the existing canonical page preserves context and prevents the same development from fragmenting into several near-duplicate URLs.

Frequently asked questions

What is NBFC AI underwriting?

NBFC AI underwriting is moving onto the Reserve Bank of India’s supervisory agenda. Deputy Governor S C Murmu urged non-bank lenders to strengthen underwriting, stress testing and risk management while using AI and machine learning carefully.

Which claims need caution?

The remarks are supervisory guidance, not a finding that AI automatically improves loan quality. Lenders remain responsible for explainability, fair treatment and model risk.

What should organisations measure?

Measure baseline cost, reliability, error rate, human review, customer impact and the evidence needed to stop or expand the deployment.

Key takeaways

• NBFC underwriting standards means the checks lenders use before approving loans.
• RBI Deputy Governor Shirish Chandra Murmu warned lenders against chasing growth with weaker checks.
• India’s rules place NBFCs into four layers based on size, risk and activity.
• Good checks can reduce bad loans, protect depositors and keep credit flowing.

NBFC underwriting standards means the checks a non-bank lender makes before giving a loan. RBI Deputy Governor Shirish Chandra Murmu warned that fast growth must not weaken those checks. NBFCs need to judge a borrower’s income, debt and ability to repay. That matters because poor loans can hurt firms and customers.

Why NBFC underwriting standards are back in focus

Non-banking financial companies, or NBFCs, lend money without being full-service banks. They finance vehicles, homes, small firms and consumer purchases. Many reach borrowers that banks may not serve easily, so their growth supports the wider economy.

But speed can bring danger. A lender may approve more loans by asking fewer questions or accepting weak documents. Murmu’s message was direct: growth should come from better systems, not looser risk checks.

Underwriting is the process of deciding whether a borrower can repay. It includes checking income, past repayments, existing debt and the value of any security. Security is an asset, such as a house or vehicle, that supports a loan.

“Growth is healthy only when the lender can still explain why each loan should be repaid,” is the central lesson from the RBI warning. This is also why the regulator keeps asking NBFCs to improve data, controls and board oversight.

What rules shape NBFC underwriting standards?

The RBI uses scale-based regulation for NBFCs. That term means bigger or riskier lenders face stronger rules. The framework has four layers: base, middle, upper and top.

The base layer covers smaller firms with limited risk. The middle layer includes most deposit-taking NBFCs and larger non-deposit lenders. The upper layer covers firms whose size or risk could affect the financial system.

Capital adequacy is one key safeguard. It means a lender must keep its own money as a cushion against losses. RBI rules generally require NBFCs to maintain a capital adequacy ratio of at least 15%.

Another safeguard is the liquidity coverage ratio. It means holding enough easy-to-sell assets to meet cash needs during a short period of stress. Covered NBFCs must build this buffer to 100% under the RBI’s rules.

RBI safeguard	Plain meaning	Key figure
Regulatory layers	Rules rise with size and risk	4 layers
Capital adequacy	Own funds to absorb losses	15% minimum
Liquidity coverage	Liquid assets for cash stress	100% for covered firms
Bad-loan test	Loan overdue before NPA status	90 days

A non-performing asset, or NPA, is a loan that stops earning income because the borrower has not paid. In general, a loan becomes an NPA after payments remain overdue for more than 90 days. Early warning signs can help lenders act before that point.

Key NBFC safeguardsRegulatory layers4Capital cushion15%Liquidity buffer100%NPA clock90 days

How can lenders grow without taking too much risk?

First, NBFCs need clear loan rules for each customer group. A vehicle loan should not use the same test as a small business loan. The lender should set limits for income, debt and loan size.

Second, firms need fresh information after approval. A borrower’s income can change, especially in a small business. Regular checks can spot trouble early, while digital records can help teams compare repayment patterns.

Third, boards must ask hard questions about growth. They should check whether new loans come from sound demand or aggressive sales targets. They should also track loans sold through agents and online partners.

RBI rules are only one part of the answer. Management teams must set the right tone, staff must follow the process and auditors must test the results. Readers can review the regulator’s latest guidance on the RBI website.

What does the warning mean for borrowers and investors?

For borrowers, stronger checks may mean more questions before approval. That can feel slow, but it may prevent a loan from becoming too large to repay. Customers should compare the interest rate, fees, late charges and total payment.

For investors, loan growth alone is not enough. They should watch bad-loan levels, credit costs, capital and cash reserves. A lender growing 30% a year may look strong, but weak repayment can erase that gain.

For the economy, careful lending can be slower at first. Still, it creates steadier credit because lenders are less likely to pull back after losses. The RBI’s banking and finance updates offer more context for readers tracking this sector.

The clearest takeaway is simple: NBFCs can expand safely only when their loan checks grow as fast as their loan books. Murmu’s warning asks lenders to treat sound credit decisions as the base of growth, not as a brake on it.

FAQs

What are NBFC underwriting standards?

They are the checks an NBFC uses to decide if a person or business can repay a loan.

Why did the RBI warn NBFCs?

The RBI warned that lenders should not approve riskier loans just to increase growth.

How can borrowers prepare for stronger checks?

Borrowers should keep income records, disclose existing debt and understand the full cost of the loan.