Skip to content

GRA · AI and Financial Model Risk

Spatial financeWhere capital meets hazard, exposure and transition

Financial-system transmission

A model error rarely stays inside one decision. Flawed or outdated models can misprice risk across a portfolio, and when many institutions use similar models, data or vendors, their errors can align and amplify market moves. Real-world shocks, in turn, break the assumptions models rely on. This page shows how real-system and model vulnerabilities propagate through institutions and markets, using sector-level analysis. It does not assess individual firms or models, or constitute advice.

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

How real-world change breaks models.

OriginChannelEffect
Economic shock or regime changeRelationships in training data no longer holdCredit, pricing and risk models misestimate
Climate and physical hazardsLosses outside the historical recordCatastrophe, credit and collateral models understate risk
Changes in customer behaviourData drift and concept driftFraud, credit and service models degrade
Data pipeline or vendor changeSilent changes to inputsErrors that monitoring may not detect
Adversarial behaviourFraudsters or users adapting to modelsControls bypassed, false outputs
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.
03Herding

Common models, common moves.

When many institutions use similar models trained on similar data, they reach similar decisions at the same moment: tightening credit, repricing risk or selling the same assets. That homogeneity is a channel of systemic risk that supervisors and the FSB have flagged in their work on AI in finance. Topic depth at model risk governance.

04Concentration

Providers behind the models.

  • Foundation model providers: few firms supply the base models many applications use.
  • Cloud compute: training and inference depend on concentrated infrastructure.
  • Data vendors: shared data sets propagate errors widely.
05Out of the models

How model failures spread.

Correlation

Similar models, similar errors

Shared vendors, data and techniques mean many firms can be wrong in the same way at the same time.

Procyclicality

Amplified market moves

Risk models that react to volatility can prompt many firms to de-risk together, deepening stress.

Concentration

Single points of failure

Reliance on a few AI and cloud providers creates shared operational dependencies.

06Consequences

Where the harm lands.

Model failures can produce unfair or inaccurate decisions for customers, losses and capital shortfalls for institutions, and instability in markets. Generative AI adds risks of misinformation and data leakage. These links connect model risk to systemic interconnectedness at Financial Stability and to cloud and outsourcing at Financial Cyber and Operational Resilience, both at GRA, and to the evaluation of AI systems at The Global Centre for Risk and Innovation (GCRI) on model evaluation.

See model monitoring and drift, third-party models and stress and scenario models. Back to the financial discipline overview.

07Analysis

Sector-level, aggregated, with limits.

Analysis uses public and aggregated information at sector level. It does not assess individual firms or models, recommend vendors or constitute advice. Scenarios state their assumptions and are not forecasts.

Any data from institutions is designed to be aggregated before use, under the safeguards set out in GRA’s written competition policy.

08Bronne

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.

09Vrae

Questions and answers.

Is competitor data shared?
No. Any data from firms is aggregated before use, under the safeguards set out in GRA’s written competition policy.
Why does model similarity matter for stability?
Because if many institutions use similar models, they may respond to the same signal in the same way, amplifying market moves and concentrating errors.
Is a transmission map a forecast?
No. It shows channels through which failures could spread. Scenarios are labelled as such and carry their assumptions.
VerslaeDra by

Dra bewyse vanaf hierdie bladsy by.

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
KI- en finansiële modelrisiko

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.