RBI Draft Guidance on Model Risk Management - Perspective 1 of 6
Published: 05 July 2026
By Nayakanti Prashant
3rd Gen Banker & Citizen
Lobbyist – Bengaluru
Digital Transactions Day (April
11)
Perspective 1 of 6
Why Indian Banking Needs an
Indian AI Governance Perspective
From Spreadsheets to Artificial
Intelligence—Every Banking Model Deserves Responsible Governance.
Every banking model—whether
powered by Artificial Intelligence, statistical techniques, business rules, or
even a carefully designed spreadsheet—ultimately serves one purpose: earning
and preserving customer trust.
We are different, we have a unique
Digital Stack.
A Six-Part Reflection Series
This six-part series presents
practical reflections on the Reserve Bank of India's Draft Guidance on Model
Risk Management, with a focus on strengthening AI governance in the context of
India's unique banking and digital payments ecosystem. The perspectives are
intended to contribute constructively to the ongoing public consultation while
encouraging a broader discussion on responsible innovation, operational
resilience, and consumer trust.
Estimated Reading time - 8
Minutes
The Scope: -
"An RE should apply these regulatory principles to all
models used by it, whether developed internally, sourced from third-parties, or
a combination thereof."
— Reserve Bank of India
Guidance on Regulatory Principles for Model Risk Management, 2026, Chapter
I, Section B (Applicability and Scope), Paragraph 6.
"Technology changes rapidly. Trust changes
slowly."
— Nayakanti Prashant
Opening Scene: The
Invisible Engines Behind Every Transaction
The day has barely begun.
A vegetable vendor in a small
town receives a payment through UPI. A pensioner withdraws money through an
Aadhaar Enabled Payment System (AEPS) terminal operated by a Business
Correspondent in a nearby village. A young entrepreneur pays GST through Bharat
Bill Payment System (BBPS). Elsewhere, a family pledges household gold to
secure a loan that is approved within minutes. A Self-Help Group receives
timely credit that supports livelihoods and local entrepreneurship.
Millions of such moments unfold
across India every single day.
To most customers, these are
simply banking services working as expected.
Behind the scenes, however,
countless models are continuously supporting decisions. Some estimate
creditworthiness. Others monitor transactions for unusual activity. Some
calculate operational risks, while others assist in fraud detection, treasury
operations, liquidity management, or customer service.
Some of these are sophisticated
Artificial Intelligence models.
Many are not.
Some are traditional statistical
models refined over years of banking experience. Others rely on business rules
or analytical techniques developed for specific operational needs. In certain
situations, even a spreadsheet used for material business or risk decisions may
fall within the broader governance expectations discussed in the Reserve Bank
of India's Draft Guidance on Model Risk Management.
Different technologies.
One common responsibility.
Good governance begins where
materiality begins.
Every model that materially
influences banking decisions deserves appropriate governance.
It is against this backdrop that
the Reserve Bank of India's draft guidance assumes significance. Far from being
a document intended only for Artificial Intelligence specialists, it represents
a broader framework for ensuring that models influencing financial decisions
remain transparent, reliable, accountable, and resilient.
The consultation arrives at an
important moment in India's financial journey. As digital banking becomes
increasingly embedded in everyday life, confidence in the systems making or
supporting decisions becomes just as important as the decisions themselves.
After all, customers rarely see
the model.
They experience its consequences.
📌 Perspective snapshot
"Model Risk Management is
not about regulating Artificial Intelligence alone. It is about responsibly
governing every material model that influences banking decisions."
Three reflections from the RBI
Draft Guidance
1. Every material model deserves
governance.
A model need not be based on Artificial Intelligence to warrant robust
governance. Depending on its purpose, criticality, and business impact, even a
spreadsheet supporting material banking decisions may fall within the broader
governance expectations outlined in the RBI's draft guidance.
2. Materiality matters more than
complexity.
The level of governance should be proportionate to a model's potential impact
on customers, institutions, and financial stability—not merely the
sophistication of the underlying technology.
3. India's banking ecosystem
deserves an Indian governance perspective.
With UPI, AEPS, BBPS, Business Correspondents, Aadhaar-enabled services,
Self-Help Groups, group lending, gold loans, and one of the world's most
diverse digital banking ecosystems, India has a unique opportunity to develop
AI and model governance that is globally respected and locally relevant.
Together, these three
observations form the foundation for the remaining five Perspectives in this
series.
Why This Consultation Matters
One of the most refreshing
aspects of the RBI's Draft Guidance is that it encourages readers to move
beyond the common misconception that Artificial Intelligence governance is only
about Large Language Models or Generative AI. In reality, governance begins
wherever models influence material decisions. The technology may evolve
rapidly, but the principles of accountability, validation, transparency, and
oversight remain remarkably consistent.
For many readers, the phrase
"Model Risk Management" may initially appear relevant only to large
banks, quantitative analysts, or technology teams.
In reality, its implications
extend much further.
Every institution that develops,
deploys, validates, monitors, or relies upon models to support material
decisions has a stake in the quality of model governance. This includes banks,
non-banking financial companies, payment system participants, fintech firms,
auditors, consultants, technology providers, and academic researchers working
at the intersection of finance and technology.
More importantly, the
consultation matters to every citizen who expects financial services to remain
safe, fair, inclusive, and trustworthy.
Artificial Intelligence has
understandably attracted significant public attention over the past few years.
Yet one of the strengths of the RBI's draft guidance is that it encourages
readers to look beyond the current excitement surrounding Generative AI.
The discussion is fundamentally
about models—their design, governance, validation, monitoring, accountability,
and responsible use throughout their lifecycle.
That distinction deserves
emphasis.
When most people hear the term AI
model, they immediately think of conversational systems, image generators,
or Large Language Models.
Banking is different.
A material model influencing
customer outcomes may be a machine learning system, a statistical forecasting
engine, a rules-based decision framework, or, depending on its purpose and
usage, even a spreadsheet supporting significant business decisions.
The technology may differ.
The governance responsibility
does not.
This broader perspective reflects
an important reality.
Good governance should never
depend solely on technological sophistication. Instead, it should be
proportionate to the impact a model can have on customers, institutions,
markets, and financial stability.
In other words, the central
question is not:
"Is this Artificial
Intelligence?"
The more meaningful question is:
"Can this model materially
influence a banking decision, customer outcome, operational process, or
financial risk?"
If the answer is yes, robust
governance becomes essential.
Beyond Global Frameworks: Why India Needs Its Own Perspective
Around the world, regulators,
central banks, and international standard-setting bodies are actively
developing frameworks to govern Artificial Intelligence and advanced analytical
models.
These efforts provide valuable
guidance and deserve careful study.
Yet India occupies a distinctive
position in the global financial landscape.
Few countries simultaneously
operate at India's scale, diversity, and pace of digital financial innovation.
Consider just a few examples.
India's Unified Payments
Interface (UPI) has transformed retail digital payments into an everyday
utility. Aadhaar Enabled Payment System (AEPS) has extended banking access to
remote communities through Business Correspondents. Bharat Bill Payment System
(BBPS) has created an interoperable ecosystem for recurring payments. Digital
lending, multilingual customer interactions, financial inclusion initiatives,
Self-Help Groups, group lending, gold loans, and Aadhaar-enabled authentication
collectively create an ecosystem unlike any other.
Each of these innovations
generates opportunities.
Each also introduces distinctive
governance considerations.
A model performing exceptionally
well in another jurisdiction may encounter entirely different operational
realities in India, where linguistic diversity, varying digital literacy,
intermittent connectivity, assisted banking channels, and large transaction
volumes shape customer experiences every day.
The future of AI governance
cannot simply be imported.
It must
also be informed by India's own banking realities.
India's
Banking Story Cannot Be Reduced to Algorithms
India's financial ecosystem has
never been built around technology alone.
Its enduring strength lies in the
way technology, regulation, institutions, and human relationships have evolved
together.
A farmer seeking seasonal credit,
a senior citizen collecting a pension through a Business Correspondent, a
street vendor accepting a UPI payment, a woman entrepreneur receiving credit
through a Self-Help Group, a student paying examination fees online, or a
family pledging jewellery for an emergency gold loan—all represent different
financial journeys.
Behind these journeys are equally
different models supporting decisions.
Some assess credit risk.
Some detect fraud.
Some predict liquidity
requirements.
Some estimate operational risks.
Some assist customer service.
Increasingly, some learn and
improve over time through Artificial Intelligence.
The challenge for governance is
therefore not simply managing technological complexity.
It is managing the consequences
of decisions.
A relatively simple spreadsheet
used to estimate provisioning, monitor liquidity, or support lending decisions
may have a greater operational impact than a sophisticated AI model used for an
internal pilot project.
The RBI's own definition
reinforces this broader perspective. The Guidance illustrates that a
spreadsheet-based loan pricing calculator, when used to determine lending
rates, customer margins, or credit terms, should itself be regarded as a model
because it materially influences business decisions.
This simple illustration
powerfully reminds us that responsible governance begins with impact, not
technological sophistication. (Chapter I, Section C, Paragraph
7(3) – Illustration)Likewise, a highly advanced machine learning model
operating under robust governance may present lower overall institutional risk
than an inadequately controlled manual process.
This is why the RBI's draft
guidance deserves careful reading.
Its underlying philosophy is
refreshingly practical.
Governance should be
proportionate to the model's significance—not merely to the sophistication of
the technology powering it.
That principle has the potential
to shape the next generation of responsible banking innovation in India.
India's
Scale Demands Operational AI Governance
Every nation approaches AI
governance through the lens of its own financial system.
India's lens is distinctive.
The country processes billions of
digital payment transactions every month while simultaneously serving customers
across metropolitan cities, small towns, remote villages, and geographically
challenging regions.
The same banking system must
accommodate digitally fluent customers using smartphones and first-time users
accessing financial services through assisted channels.
Some customers interact in
English.
Many do not.
Some have uninterrupted broadband
connectivity.
Others rely on unstable mobile
networks.
Some comfortably navigate digital
interfaces.
Others depend upon Business
Correspondents or family members to complete transactions.
These realities matter because
models do not operate in isolation.
They operate within environments
shaped by people, infrastructure, language, regulation, and trust.
An AI-enabled customer support
system trained primarily on English-language interactions may perform
exceptionally in one segment while struggling to understand customers
communicating in regional languages.
A fraud detection model may need
to distinguish between genuinely suspicious behaviour and legitimate
transaction patterns emerging from seasonal agricultural activity.
A credit assessment model
developed for salaried urban borrowers may require careful adaptation before
being deployed for Self-Help Groups, micro-enterprises, or agricultural
lending.
Similarly, models supporting
Aadhaar-enabled authentication or assisted banking environments may need to
account for operational realities that differ significantly from purely
app-based customer journeys.
These are not merely technical
questions.
They are governance questions.
Responsible AI governance begins
by recognising that context matters.
Trust:
The Currency That Cannot Be Digitised
Digital transactions may travel
in milliseconds.
Trust does not.
Trust is accumulated
gradually—through consistent experiences, transparent decisions, effective
grievance redressal, and confidence that technology operates fairly even when
customers do not fully understand how it works.
Customers rarely ask whether a
decision originated from Artificial Intelligence, a statistical model, or a
spreadsheet.
They ask simpler questions.
"Why was my transaction
declined?"
"Why was my loan application
rejected?"
"Why was additional
verification required?"
"Why did this recommendation
suddenly change?"
Every unanswered question becomes
an opportunity for trust to weaken.
Every well-governed model becomes
an opportunity for trust to strengthen.
That is why explainability,
documentation, validation, accountability, and human oversight are no longer
optional governance practices.
They are essential components of
customer confidence.
The future of digital banking
will be determined not only by how intelligent our models become, but by how
confidently institutions can explain, govern, and improve them.
AI
Governance Is a Collective Responsibility
Artificial Intelligence is often
portrayed as the domain of software engineers and data scientists.
That perception is incomplete.
The governance of models
influencing banking decisions extends far beyond technology teams.
It requires collaboration across
multiple disciplines.
Risk managers evaluate
institutional exposure.
Auditors assess controls.
Lawyers interpret legal
obligations and evolving regulatory expectations.
Compliance professionals ensure
adherence to supervisory requirements.
Behavioural scientists and
selected psychologists studying human–AI interaction help us understand
automation bias, customer behaviour, decision fatigue, and the subtle ways in
which people place trust—or lose it—in intelligent systems.
Policymakers shape the broader
regulatory environment.
Bankers contribute operational
experience developed through decades of serving diverse customer segments.
Ultimately, every citizen
benefits when these perspectives converge to build trustworthy financial
systems.
As India's banking ecosystem
continues its remarkable digital transformation, one principle deserves careful
reflection:
The future of AI governance will
not be written by technologists alone. It will be shaped equally by bankers,
regulators, auditors, lawyers, behavioural scientists, policymakers, and
citizens who collectively define what trustworthy banking should look like.
Perhaps that is the most
significant opportunity presented by the RBI's Draft Guidance on Model Risk
Management.
It
invites us to view governance not as a compliance exercise, but as a shared
commitment to preserving public trust in an increasingly intelligent financial
system.
An
Opportunity Beyond Compliance
Every major technological
transformation eventually reaches a defining moment.
For electricity, it was safety
standards.
For aviation, it was operational
discipline.
For digital payments, it was
trust built through regulation, interoperability, security, and continuous
innovation.
Artificial Intelligence and
model-driven banking have now reached a similar stage.
The Reserve Bank of India's Draft
Guidance on Model Risk Management should therefore be viewed not merely as
another regulatory document, but as an opportunity to strengthen one of the
most valuable assets in banking—public confidence.
Good governance does not slow
innovation.
It enables sustainable
innovation.
Banks that invest in robust model
governance today are likely to be better prepared for tomorrow's technological
advances, regulatory expectations, and customer needs. Well-governed models can
improve decision-making, reduce operational surprises, strengthen consumer
protection, and enhance institutional resilience.
Equally important, good
governance recognises that every model has a lifecycle.
Models are designed.
They are tested.
They are deployed.
They evolve.
Sometimes they perform
exceptionally.
Sometimes they drift from their
intended behaviour.
Occasionally, they should be
modified, temporarily suspended, or retired altogether.
Preparing for these possibilities
is not a sign of mistrust in technology.
It is a hallmark of responsible
governance.
This perspective becomes
particularly relevant as banks increasingly rely upon third-party models,
cloud-based AI services, external APIs, and rapidly evolving Artificial
Intelligence capabilities.
Responsible institutions do not
merely ask:
"Can this model be
deployed?"
They also ask:
"How will we monitor
it?"
"Who remains
accountable?"
"When should it be
independently reviewed?"
"Under what circumstances
should it be modified, suspended, or retired?"
These questions are not obstacles
to innovation.
They are the foundations of
trustworthy innovation.
A
Beginning, Not a Conclusion
This first perspective has
intentionally focused on the broader landscape.
Before discussing specific
governance mechanisms, it is important to recognise why India's banking
ecosystem deserves an Indian approach to model governance.
The conversation is larger than
Artificial Intelligence alone.
It encompasses every material
model that influences financial decisions—whether built using advanced machine
learning, traditional statistical techniques, business rules, or even
spreadsheets supporting significant business processes.
Technology may continue to
evolve.
Governance principles endure.
As India's digital economy grows,
the measure of success will not simply be the intelligence of our models.
It will be the confidence with
which customers, regulators, institutions, and society trust the decisions
those models help produce.
That confidence cannot be coded
into software.
It must be earned through
thoughtful governance.
Looking
Ahead
This article has explored why
India's unique banking ecosystem requires an equally distinctive perspective on
model governance.
The next article in this series
turns to the equally important question:
What are the principal
operational risks that banks should prepare for as Artificial Intelligence and
analytical models become increasingly integral to financial services?
Rather than examining these risks
solely through a technological lens, the discussion will focus on their
practical implications for governance, consumer protection, operational
resilience, and institutional accountability.
Digital Transactions Day Reflection
Every secure digital transaction
is built on an invisible foundation of trust—trust in the technology, trust in
the institution, and trust in the governance that protects both. As India
continues to lead the world in digital payments and financial innovation,
strengthening the governance of AI and analytical models is a natural extension
of that trust.
This reflection is also shared in
support of the proposed observance of April 11 – Digital Transactions Day,
commemorating the launch of the Unified Payments Interface (UPI) pilot in 2016.
The proposal seeks to encourage greater awareness of secure, inclusive, and
responsible digital transactions for every citizen.
Disclaimer
This article represents the
author's personal reflections on the Reserve Bank of India's Draft Guidance on
Model Risk Management. The views expressed are intended solely to contribute
constructively to the ongoing public consultation and do not represent the
views of any employer, institution, regulator, or affiliated organisation.
References to the proposed April
11 – Digital Transactions Day are part of the author's long-standing public
awareness initiative to promote safe, inclusive, and trusted digital
transactions.
References
1.
Reserve Bank of India – Draft Guidance on Model
Risk Management (Public Consultation).
2.
Reserve Bank of India – Report of the FREE-AI
Committee (Framework for Responsible and Ethical Enablement of Artificial
Intelligence in the Financial Sector).
3.
Bank for International Settlements (BIS) –
Publications on Artificial Intelligence, Machine Learning, Model Risk, and
Digital Innovation in Banking.
4.
Financial Stability Board (FSB) – Artificial
Intelligence and Financial Stability publications.
5.
National Payments Corporation of India (NPCI) –
UPI, AEPS, BBPS, and related payment ecosystem resources.
6.
Relevant publications issued by international
standard-setting bodies and academic institutions on Model Risk Management,
Artificial Intelligence governance, consumer protection, operational
resilience, and financial inclusion.
"The strength of a financial
system is measured not only by the intelligence of its models, but by the
wisdom of its governance."
— Nayakanti Prashant
Coming Next
Perspective 2 of 6
Beyond Principles:
Seven Operational AI Risks Banks Should Prepare For
The Joy of Digital Transactions
Nayakanti Prashant
3rd Gen Banker & Citizen Lobbyist – Bengaluru
Digital Transactions Day (April 11)
Author’s Blogs
https://prashantrandomthoughts.blogspot.com
https://prashantnepayments.blogspot.com
https://innovationinbanking.blogspot.com
Comments
Post a Comment