Climate risk is moving from voluntary reporting to mandatory compliance for Indian banks on a defined timeline. The convergence of RBI's mandatory climate disclosure framework, CRIS data infrastructure, and growing green finance portfolios means banks are deploying AI for climate risk work — and those AI models fall under RBI MRMF's governance scope. The intersection of climate risk governance and AI model governance is a new compliance challenge that most Indian banks' risk teams are approaching as two separate workstreams. They are not separate.
An AI model that calculates climate physical risk exposure for a bank's loan portfolio is a consequential model under MRMF: its output directly feeds the mandatory regulatory disclosure. An AI model that adjusts credit scores for ESG factors is a credit model under MRMF, regardless of what data it uses. A climate stress testing model that projects portfolio losses under a 2°C or 4°C scenario is used for capital planning — a high-materiality AI use case. All three require model inventory, validation, monitoring, and human oversight documentation.
RBI's Phased Climate Disclosure Framework: Mandatory Timeline
Governance, Strategy, Risk Management
- Board and senior management oversight of climate-related risks and opportunities
- How climate risk affects the institution's business model, strategy, and financial planning
- Risk management processes for identifying, assessing, and managing climate risk
- Integration of climate risk into overall enterprise risk management framework
- Disclosure of how AI tools are used (where applicable) for climate risk assessment
Quantitative Climate Risk Metrics and Targets
- Portfolio-level climate physical risk exposure by sector and geography
- Financed emissions (Scope 3 Category 15) measurement and reporting
- Portfolio alignment with net-zero transition pathways
- Climate-related concentration risk by sector and counterparty
- Progress against climate risk management targets over time
FY2026 climate disclosure is already mandatory. Banks reporting on climate risk governance without AI model governance documentation for their climate risk AI tools have a gap: if the disclosure describes how AI is used for climate risk assessment (which IFRS S-1 transparency requirements suggest it should), and the AI has no MRMF governance, the disclosure itself creates regulatory exposure. The governance program and the disclosure program must be co-developed.
Five ESG AI Use Cases in Indian Banking and Their MRMF Status
Climate Stress Testing AI
AI models that project portfolio performance under climate scenarios (physical risk: extreme weather events; transition risk: carbon tax, stranded assets). Used in capital planning and regulatory submissions. High-materiality under MRMF — errors affect capital adequacy assessments. MRMF requirements: rigorous validation against historical climate data, scenario methodology documentation, uncertainty range disclosure, expert human review before capital planning use. Climate stress testing AI is among the highest-materiality AI in banking after credit decisioning AI.
Green Credit Risk Scoring AI
AI that incorporates ESG ratings into credit decisions — adjusting credit scores based on borrower's environmental rating, sector transition risk, physical risk to collateral. ML-based ESG credit models reduce default prediction error by up to 23% vs traditional models, but require MRMF validation. Key governance risks: ESG data quality is lower for SMEs and emerging sectors; ESG data proxies may correlate with unintended characteristics; ESG score weighting in credit decisions must be explainable to borrowers under RBI MRMF explainability requirements. Borrower who receives adverse credit decision based on ESG factors must be able to receive a meaningful explanation.
CRIS-Integrated Physical Risk AI
AI that uses India's Climate Risk Information System (CRIS) geospatial data to assess physical climate risk for loan collateral: flood zone classification, heat stress risk, drought exposure for agricultural loans, cyclone risk for coastal collateral. CRIS data quality and coverage limitations must be documented in MRMF governance — CRIS geographic coverage may be incomplete for some regions, and banks must understand where CRIS data is absent and how the AI handles missing data. CRIS-integrated AI with undocumented data gaps produces unreliable physical risk assessments in exactly the regions where physical risk may be highest.
Financed Emissions Calculation AI
AI models that calculate a bank's Scope 3 Category 15 financed emissions — the emissions attributable to the bank's loan and investment portfolio. FY2028 MRMF metrics requirement will mandate this disclosure. Banks preparing now must govern their financed emissions AI: methodology documentation (PCAF standard or equivalent), data source documentation (borrower-reported vs estimated emissions), uncertainty quantification, and validation against sector emissions benchmarks. Financed emissions AI with undocumented methodology cannot be included in credible regulatory disclosure.
Green Taxonomy Classification AI
AI that classifies financial products as green under India's developing green taxonomy (aligned to EU Taxonomy and RBI green deposit framework). Greenwashing risk: AI that incorrectly classifies non-green products as green creates regulatory and reputational exposure. MRMF governance for taxonomy AI: classification methodology documentation (which criteria define 'green' for each product category), validation against RBI green deposit framework, monitoring for taxonomy evolution (as India's green taxonomy develops, classification AI must be updated and re-validated), and audit trail for every classification decision supporting green product disclosure.
Portfolio Alignment AI
AI that assesses a bank's loan portfolio alignment with India's NDC (Nationally Determined Contribution) commitments and net-zero transition pathways. Portfolio alignment AI projects how the portfolio's sector composition and counterparty characteristics align with India's decarbonization trajectory to 2050. Governance requirements: sector transition pathway documentation (what trajectory does the AI assume for each sector), counterparty data quality assessment, and sensitivity analysis showing how different transition scenarios affect alignment scores. Portfolio alignment AI outputs inform strategy disclosures under RBI's Phase 1 mandatory framework — they must be defensible under regulatory examination.
Power Sector, Infrastructure Banks, Agricultural Credit AI, and Priority Sector Climate Risk
- Renewable energy lending AI: India's renewable energy sector is the largest recipient of green finance: solar, wind, and hydro project loans. Banks providing renewable energy project finance use AI for project cash flow modeling, asset performance monitoring, and early warning systems for project underperformance. MRMF governance for renewable energy AI must address: project-specific performance data integration (satellite-based solar irradiance data, wind resource data), model calibration for India's electricity market structure (DISCOM payment risk, grid curtailment risk), and documentation of how AI distinguishes between renewable energy underperformance due to weather (climate risk) versus operational failures (counterparty risk).
- Agricultural credit AI and climate physical risk: India's priority sector lending (PSL) mandate directs a significant portion of bank credit to agriculture. Agricultural credit AI faces India-specific climate physical risk challenges: the same geography may have high drought risk for kharif crops and high flood risk for rabi crops. AI models for agricultural credit must incorporate crop-specific and season-specific climate risk, not just geographic risk. The Kisan Credit Card (KCC) program and the PM-KISAN farmer database represent data infrastructure for agricultural credit AI — CRIS integration with agricultural credit AI creates specific governance obligations for rural lending AI across India's 640+ districts.
- Infrastructure lending and transition risk AI: India's infrastructure pipeline — highways, airports, ports, urban infrastructure — is the largest in Asia outside China. Banks financing long-tenor infrastructure projects face transition risk: infrastructure built for a high-carbon economy may become stranded in a net-zero transition. AI models that assess transition risk for infrastructure loans must project policy trajectories (carbon pricing, fuel efficiency standards, building codes) across loan tenors of 15-25 years. The uncertainty over 25-year policy trajectories requires MRMF-documented uncertainty quantification and sensitivity analysis for transition risk AI in infrastructure lending.
- SEBI's ESG disclosure requirements and fund manager AI: SEBI's Business Responsibility and Sustainability Reporting (BRSR) framework and ESG disclosure requirements for listed companies and mutual funds create ESG data that fund managers use in investment AI. SEBI-regulated asset managers deploying AI for ESG screening, impact investing, and sustainable fund portfolio management are subject to SEBI's model governance expectations. The RBI MRMF and SEBI's governance frameworks are converging on the same ESG AI use cases from different regulatory angles — fund managers serving both equity markets (SEBI) and lending institutions (RBI) face dual governance obligations for their ESG AI platforms.