The rapid rise of Artificial Intelligence (AI) in banking is transforming how financial institutions operate. From customer service chatbots to credit scoring, fraud detection, and new product design, AI promises efficiency, scalability, and deeper outreach.

But as the Reserve Bank of India (RBI) Deputy Governor M. Rajeshwar Rao has emphasised, this must be tempered by responsible adoption. This article explores RBI’s views, the framework it has launched (FREE-AI), the challenges involved in deploying AI in the financial sector, and how Indian banks can navigate risks while capturing opportunities.

AI in Banking: Opportunities and Growing Adoption

AI is no longer futuristic in Indian banking; it is already being used in core operations.

According to surveys by RBI in 2023-24, more than three-fourths of Indian banks have deployed AI-powered chatbots for customer service.
AI is being used for credit scoring (including for those without traditional credit history), fraud detection, document verification, and customised credit or product offers.

These applications can lead to lower costs, faster services, greater financial inclusion, and improved risk management. However, with high promise comes high risk, especially in a sector handling people’s money and trust.

RBI’s Perspective: Measured, Responsible, Ethical Use

At a recent CNBC-TV18 Banking Transformation Summit in Mumbai, Deputy Governor M. Rajeshwar Rao underlined that:

AI adoption in banking must be measured and responsible. It should not be treated simply as a technological upgrade, but as a structural shift affecting all levels of banking operations.
The excitement around benefits, profitability, efficiency, competitiveness, should not overshadow prudent risk management and long-term stability.
Key elements include strong governance, ethical use, fairness, transparency (e.g. explainability), and alignment with stability and resilience of the financial system.

He also noted that an RBI study found a sharp increase in references to AI in banks’ annual reports, reflecting its rising strategic importance.

RBI’s Regulatory Response: FREE-AI Framework

To address both promise and peril, RBI constituted the FREE-AI Committee (Framework for Responsible and Ethical Enablement of Artificial Intelligence) which released a report on 13 August 2025.

Key Components of FREE-AI

Seven guiding principles (“sutras”), including:
- Trust is the foundation
- People first
- Innovation over restraint
- Fairness and equity
- Accountability
- Understandable by design
- Safety, resilience and sustainability
Six strategic pillars that group actionable recommendations under two broad themes: innovation enablement and risk mitigation. These are

Innovation enablement pillars:
- Infrastructure
- Policy
- Capacity
Risk mitigation pillars:
- Governance
- Protection
- Assurance

Some Specific Recommendations

Developing indigenous models that better reflect India’s linguistic, cultural, and socio-economic diversity rather than depending wholly on general or foreign AI systems.
Establishing AI innovation sandboxes for experimentation in controlled settings.
Ensuring regulated entities have board-approved AI policies, strong internal governance, audit mechanisms, oversight, and risk frameworks.
Clear rules or guidelines for dealing with AI model risks: bias, transparency, explainability, data quality, third-party dependencies, vendor risks, cybersecurity, etc.

Challenges & Risks of AI Adoption in Banking

While the FREE-AI framework addresses many concerns, the implementation will be nontrivial. Key challenges include:

Model bias, opacity, and fairness issues: AI systems trained on historical data may perpetuate discrimination or bias. If decision pathways are not explainable, customers may not understand, nor get recourse for, why a decision was made.
Data risks and privacy: The quantity, quality, and lineage of data are critical. Fragmented, outdated, incomplete or unrepresentative data could lead to wrong or unfair outcomes. There’s also risk of data breaches or misuse.
Vendor / third-party risk and concentration risk: Relying heavily on external AI vendors or few providers leads to systemic dependencies. A failure in one vendor might affect many banks.
Model drift, reliability, security: Over time, models may degrade (due to changing environments), adversarial attacks may compromise models, and generative AI may be misused (deepfakes etc.).
Regulatory and capacity challenges: Banks, smaller NBFCs, cooperative banks often lack sufficient AI expertise. Regulators also need sufficient technical capacity. Laws may not yet fully address AI-specific issues.

Benefits vs Risks

Below is a table summarising prominent benefits of AI adoption in banking alongside the corresponding risks, to help understand the trade-offs clearly.

Key Benefits vs Associated Risks of AI in Banking

Benefit	Key Risk(s)
Enhanced customer service (e.g. chatbots, 24×7 support)	Misinterpretation, biased responses, inability to handle edge cases
Improved fraud detection & risk analytics	False positives/negatives, adversarial attacks, system vulnerabilities
Better credit inclusion through alternative data	Bias in alternative data, lack of explainability, unfair exclusion
Operational efficiency & cost reduction	Over-automation, loss of human oversight, reliance on opaque systems
Product customisation & competitive differentiation	Privacy infringement, model misuse, regulatory non-compliance

This table shows that for each benefit, there is a counterbalancing risk. The impact of risks can range from minor customer dissatisfaction to serious regulatory, legal, or reputational damage. Thus, any institution aiming to adopt AI must invest effort and resources in the risk side just as much as in seizing benefits.

Implementation & Governance: What Banks Need to Do

To harness AI responsibly, Indian banks (and financial entities) should follow a roadmap that implements the recommendations within the FREE-AI framework. Some of the critical steps are:

Establish board-level ownership of AI strategy: policy, risk appetite, oversight. Governance should ensure alignment with core principles (fairness, accountability, etc.).
Institute explainability and transparency: ensure AI decisions (e.g. credit rejection) can be explained to customers; allow for challenge or override where needed.
Robust vendor management and contractual safeguards: require third-party AI providers to adhere to standards around bias, security, audit rights, data usage.
Continuous monitoring, audit, testing: models need to be validated, retrained, monitored for drift. Also stress-tested under various conditions.
Capacity building: human skills in AI risk, data science, cyber-security; regulatory training; hiring or upskilling staff.

FREE-AI Framework Pillars & Recommendations

Here is a table showing major pillars of the FREE-AI framework and some of the specific recommendations under each pillar. This gives a more concrete view of what the regulatory expectations are.

FREE-AI Framework: Pillars & Example Recommendations

Pillar	Example Recommendations
Infrastructure	Creation of India-specific data platforms; digital public infrastructure to support model development; fostering indigenous AI models.
Policy	Establish AI innovation sandboxes; adaptive policies; dedicated funding for India-relevant models.
Capacity	Training for boards, C-suite; technical workforce development; regulators’ own capacity enhancements.
Governance	Board-approved AI policies; ethical governance; risk-appetite frameworks; audit / oversight mechanisms.
Protection	Consumer protection rules; data privacy and consent; bias mitigation; transparency in outcomes.
Assurance	Audit trails; independent audits; incident reporting; standards for security, safety, resilience.

The table above clarifies how the FREE-AI framework relates high-level principles to concrete actions. It demonstrates that governance is not an afterthought, but integral at every stage (design, deployment, oversight).

Global & Comparative Context

India is not alone in trying to balance AI innovation with risk and regulation. Other jurisdictions provide useful comparisons.

The EU’s Artificial Intelligence Act (2024) is one of the more comprehensive laws, classifying AI systems by risk category and imposing corresponding obligations.
The UK, Singapore and the U.S. tend to favour guidance and risk-based regulation rather than prescriptive rules, especially for early stage or lower risk use cases.

India’s approach via FREE-AI is somewhere in the middle: principle-based, somewhat prescriptive in places (e.g. governance, audits), but also allowing space for innovation through sandboxes and leniency (e.g. first errors).

Potential Impacts & Long-Term Implications

Financial inclusion: Well-designed AI systems could significantly reduce barriers for underbanked populations, credit invisibles, by using alternative data. But missteps could entrench exclusion or unfair treatment.
Trust & Reputation: Banking relies heavily on trust. If AI led to unfair outcomes, data breaches, or opaque decisions, public trust could erode, which is costly and hard to regain.
Systemic stability: Concentration risk, vendor failures, model “herding” effects (where many banks using similar models behave similarly) could pose systemic threats.
Regulatory burden and cost: Complying with strong governance, audits, transparency, vendor oversight adds cost; smaller banks or NBFCs may struggle unless support is provided.

Conclusion

India stands at a critical juncture in integrating artificial intelligence into its banking sector. On one hand, AI offers transformative potential: service delivery, inclusion, efficiency. On the other hand, it gives rise to risks, bias, opacity, security, systemic vulnerabilities, that if ignored, could undermine trust, fairness, and stability.

The RBI, through Deputy Governor M. Rajeshwar Rao’s exhortations, and through the FREE-AI framework, has made it clear that innovation must proceed hand in hand with strong governance, ethical design, transparency, and human oversight.

For banks, fintechs, NBFCs and regulators, the path ahead will require balancing opportunity with caution, investing in capacity and infrastructure, and ensuring that AI works for people, and not at their expense.

Other News Pages
Bank of Baroda Diwali Rewards & Digital Offers	Banking Basics Every Banker Should Know	Tata Capital to Expand Used Vehicle Loans	AI in Banking — Can Robots Cheat?
AD Cat-I Banks Now Allowed in India	RBI: Stop Comparing the Rupee with USD	Banks Borrow ₹1 Lakh Crore Overnight from RBI	Japan Upgrades India’s Long-Term Credit Rating
RBI Urges Banks to Cut Loan & Card Fees	Fewer Opportunities to Study Abroad	RBI to Review Rules with Expert Panel	GST Cut from Sept 22 to Lower Daily Costs
SBI Pushes for Universal Banking in Tamil Nadu	Banks Use Robots to Recover Loan Dues	RBI: No Phone Lock for Loan Defaults	8 Years After GST — Double Benefit for Middle Class