Skip to content

GRA · AI and Financial Model Risk

Spatial financeWhere capital meets hazard, exposure and transition

Financial discipline overview

Model risk management is the discipline of controlling the risks that arise from using quantitative models and AI systems in financial decisions. It developed in banking around valuation and capital models and now covers credit, insurance pricing, reserving, trading, fraud detection, anti-money-laundering, customer service and generative AI tools. This page describes the discipline as research rather than execution: where models are used, how they are governed and what shared understanding the industry needs.

STEM methods
4
Canonical methods this field applies
Topics
8
Subject pages in the field
Counterparts
5
Platforms in shared fields across the network
Uchungechunge lwemibiko
2
Reports series it contributes to
01Where models are used

The model landscape in finance.

AreaTypical modelsMain risks
LendingCredit scoring, probability of default, affordabilityUnfair outcomes, poor performance in new conditions
InshuwalensePricing, reserving, catastrophe and claims modelsMispricing, reserve shortfalls, discrimination
IzimaketheValuation, trading, market riskMis-valuation, correlated behaviour in stress
Capital and liquidityRegulatory and economic capital, stress testingUnderstated risk, false comfort
Ubugebengu bezezimaliFraud and transaction monitoringMissed crime, excessive false alerts
Operations and serviceGenerative AI assistants, document processingInaccurate output, data leakage, dependency
02Monitor

Field overview.

The screen a team would use to follow this subject: its measures, states and decisions. Names are examples and values are illustrative.

Field overview: the screen’s structure, terms and states for AI and model risk.Example names and illustrative values; a design, not a record or a live system.
03Model uses

Where models drive financial decisions.

SebenzisaModel typeMain risk
Credit decisionsScorecards and machine learningBias, drift
Insurance pricing and reservingGLMs, machine learning, catastrophe modelsNon-stationarity, opacity
Capital and stress testingInternal models, scenario modelsAssumption risk
Fraud and AMLAnomaly detectionFalse positives and negatives
Operations and customer serviceGenerative AIHallucination, data leakage

Shocks travel between these parts, and AI and Financial Model Risk studies the links: When does a model fail, and who notices? The evidence base is model records and monitoring data.

04How models are governed

The core of the discipline.

  1. 01

    Inventory and tiering

    Knowing every model in use, its owner and its materiality. See model risk governance.

  2. 02

    Amazinga entuthuko

    Documented design, data and assumptions, with testing before use.

  3. 03

    Independent validation

    Effective challenge by people independent of the developers.

  4. 04

    Ukuqapha

    Ongoing checks of performance and drift, with thresholds for review. See model monitoring and drift.

  5. 05

    Use controls

    Human oversight and limits on where a model may be used. See human-in-the-loop controls.

05Research, not execution

What shared understanding the industry needs.

Each institution governs its own models. What none can see alone is how model risk accumulates across the sector: how many firms rely on similar data, vendors or techniques, how methods for validating machine learning and generative AI compare, and where common failures are emerging. That precompetitive picture is the focus of this platform. How model failures can spread is set out on financial-system transmission.

Methods include uncertainty quantification and model validation, AI evaluation and assurance engineering, machine learning and statistics. Sector scenarios appear on the scenarios page as they are reviewed.

Nothing on this platform validates, approves or certifies any model, vendor or institution, and nothing is legal, regulatory or investment advice. Analysis is sector-level research.

06Imithombo

Where this page’s references lead.

The bodies and documents this page is read against. Status and editions change; each link goes to the issuing body.

07Pages

Pages in Financial discipline overview.

08Imibuzo

Questions and answers.

Does GRA advise firms on their models?
GRA is a nonprofit business league for the financial services industry. It does not lend, underwrite, rate, invest, advise on or arrange transactions, and readiness work is not approval.
Is model risk only a banking concern?
No. It applies wherever models inform financial decisions, including insurance, asset management, payments and fintech, and increasingly where generative AI is used in operations.
Why does AI change model risk management?
Because machine learning models can be numerous, frequently retrained, harder to explain and more sensitive to shifts in data, and generative AI can be adopted informally across an institution. Frameworks built for a smaller number of statistical models must adapt their inventories, validation methods and monitoring.
Which methods does the discipline use?
Artificial intelligence and machine learning, AI evaluation and assurance engineering, uncertainty quantification and model validation, and probability and statistics. See the STEM foundation page.
ImibikoNikela

Nikela ngobufakazi kusuka kuleli khasi.

Reports is the Consortium’s quarterly publication. What you know about this subject can become evidence, a case or a correction in the next edition, credited to you and published under CC BY 4.0 where it is your own work.

Global Risks AllianceAI and Financial Model RiskMethods and TechnicalExponential Technology
Ubungozi Bamamodeli e-AI Nawezezimali

Bring an exposure or capital question to the Alliance.

Information and enquiries are open today. An enquiry receives a reference and names the next step; applications, payments, compute allocations and registrations open only when their team, terms and review capacity exist.