GRA · AI and Financial Model Risk
Spatial financeWhere capital meets hazard, exposure and transition
AI and Financial Model Risk
Model governance and responsible AI use in financial decision processes and risk measurement. AI and Financial Model Risk is an established platform of The Global Risks Alliance.
- 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
AI and Financial Model Risk in seven parts.
- 01FrameworksWhat the work is read against
- 02What the platform doesShared understanding of model risk, not validation
- 03OverviewWhat AI and Financial Model Risk does
- 04The problemModels are now part of the financial system’s plumbing
- 05Foundation and counterpartsGrounded in method, linked to science and governance
- 06CoverageThe areas of artificial intelligence this platform takes on
What the work is read against.
The platform does not set standards. It reads its evidence against the frameworks that competent bodies have adopted, so a finding can be placed where a decision-maker already works.
| Framework or standard | Body | What it sets |
|---|---|---|
| Supervisory Guidance on Model Risk Management (SR 11-7) | Board of Governors of the Federal Reserve System and Office of the Comptroller of the Currency | Expectations for model development, validation and governance in United States banks, including effective challenge |
| Model risk management principles for banks (SS1/23) | Prudential Regulation Authority (United Kingdom) | Principles for model identification, governance, development, validation and risk mitigants, covering AI and machine learning models |
| EU Artificial Intelligence Act | European Union | Risk-based obligations, under which AI used for creditworthiness assessment and for life and health insurance pricing is classed as high-risk |
| Application Paper on the supervision of artificial intelligence | International Association of Insurance Supervisors (IAIS) | How the Insurance Core Principles apply to insurers’ use of AI: governance, robustness, transparency, fairness and redress |
| Report on the financial stability implications of artificial intelligence | Financial Stability Board | Assessment of AI-related vulnerabilities such as third-party concentration, market correlation, cyber risk and model risk |
| Principles to promote Fairness, Ethics, Accountability and Transparency (FEAT) | Monetary Authority of Singapore | Principles for the responsible use of AI and data analytics in financial services |
| AI Risk Management Framework | US National Institute of Standards and Technology (NIST) | Govern, map, measure and manage functions for AI risk, with a profile for generative AI |
| Equal Credit Opportunity Act and Regulation B adverse action requirements | United States Congress; Consumer Financial Protection Bureau | Duties to give applicants specific reasons for credit denial, which apply when complex models are used |
Model risk lifecycle.
The screen a team would use to follow this field: its measures, states and decisions. Names are examples and values are illustrative.
Model risk lifecycle
Model inventory · Committee 3
Lifecycle
- Inventoryregistered
- Developdocumented
- Validateindependent
- Approvecommittee
- Implementcontrols
- Monitorthresholds
- Reviewannual
- Retiredecommission
Models
| MD-620 PD retail | Tier 1 | 19 months | 1 | Overdue |
| MD-007 LLM complaints triage | Tier 1 | never | 0 | Not validated |
| MD-256 IFRS 9 staging | Tier 1 | 10 months | 1 | Approved |
| MD-852 marketing propensity | Tier 3 | 10 months | 0 | Approved |
Validation scope
- Conceptual soundness
- In progress
- Outcomes analysis
- Pending
- Benchmarking
- Pending
- Use limits
- human review required
The questions this field works on.
Model risk governance
Inventory, ownership and validation.
TopicCredit and pricing models
Finance-specific automated decisions.
TopicGenerative AI in finance
Use cases and failure controls.
TopicExplainability and fairness
Interpretation, bias and accountability.
TopicModel monitoring and drift
Changes over time and review thresholds.
TopicHuman in the loop controls
Authority and approvals.
TopicThird party models
Vendor dependencies and due diligence.
TopicStress and scenario models
Verification and uncertainty.
When a run, a panic or an adoption tips.
The same threshold logic behind bank runs, herding and the adoption of new products: below a critical point nothing spreads, above it nearly everyone moves. Teaching model with stated illustrative values, not a prediction.
Controls
- Mean threshold (share of neighbours)
- 18%
- Spread of thresholds
- 8 points
- Early movers
- 2.0% of people
- Contacts per person
- 8
Readouts
- Share who acted in the end
- 100%
- Global cascade
- Yes
- Steps to settle
- 13
Numbers, methods and sources
| Step | Share who have acted (%) |
|---|---|
| 0 | 1.3 |
| 1 | 4.5 |
| 2 | 11.3 |
| 3 | 20.3 |
| 4 | 29.3 |
| 5 | 41.3 |
| 6 | 53.3 |
| 7 | 64.5 |
| 8 | 79.0 |
| 9 | 91.5 |
| 10 | 96.5 |
| 11 | 99.8 |
| 12 | 100.0 |
| 13 | 100.0 |
Methods
- Threshold models of collective behaviour Granovetter (1978), American Journal of Sociology 83
- Global cascades on random networks Watts (2002), PNAS 99
- Complex contagions Centola and Macy (2007), American Journal of Sociology 113
- One-at-a-time sensitivity (tornado) Saltelli and others (2008), Global Sensitivity Analysis: The Primer, Wiley
Data
No observed data enters this model: it runs on the teaching values set by the controls, to be replaced with a programme’s own evidence.
Across the six pillars, led by Nexus Consortium
- Experiments Labs
- Built in AI and Financial Model Risk in Foundry
- Tested at AI and Financial Model Risk in Universe
- Codified in Digital, Cyber and AI in Standards
- Published in AI and Financial Model Risk in Reports
- Taught in Faculty of Technology and Digital Systems in Academy
Shared understanding of model risk, not validation.
The platform is designed to give the industry a precompetitive evidence base on how financial models and AI systems are inventoried, tested, validated and monitored, and on where they fail. Each institution validates and governs its own models under its own supervisors.
Inventory, testing and monitoring
Model inventories, testing, validation and monitoring methods, compared across practice.
Methods and research
Model governance methods and AI-in-finance research, published once reviewed.
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. 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.
What AI and Financial Model Risk means for exposure, loss and capital.
The exposure, loss, scenario and capital views a risk officer, supervisor or ministry of finance would ask for, with the frameworks they report against. Structures follow supervisory and actuarial practice. Values are illustrative: never prices, ratings or advice.
Transmission channels
How a shock in AI and Financial Model Risk reaches credit, market, underwriting, liquidity and operational risk, and feeds back.
Scenario pathways
How AI and Financial Model Risk could be stressed under the scenario families central banks and supervisors use.
| Scenario family | Transition risk | Physical risk | What a stress test asks |
|---|---|---|---|
| Orderly | Early, gradual policy | Lower | Can the book absorb a steady repricing? |
| Disorderly | Late, abrupt policy | Lower to medium | What breaks under a sudden repricing and credit migration? |
| Hot house world | Limited policy | High and rising | Where do physical losses concentrate in collateral and insurance? |
What AI and Financial Model Risk does.
AI and financial model risk is the risk of loss or harm from models, including machine learning and generative AI, that are wrong or misused in finance. Global Risks Alliance’s platform explains model governance, credit and pricing models, generative AI, explainability and fairness, drift, human oversight, vendor and stress models. It validates nothing.
Its community
Model risk teams; quantitative validators; responsible AI leads.
Its methods
Model inventories; testing; validation; monitoring.
Its outputs
Model governance methods; AI-in-finance research.
Models are now part of the financial system’s plumbing.
A model is a simplified representation of reality used to make a decision: a credit score, an insurance price, a capital calculation, a fraud alert, a liquidity forecast. Every model rests on assumptions and on data drawn from the past, and every model can be wrong in two ways: it can be built or implemented incorrectly, or it can be used where its assumptions no longer hold. Model risk is the potential for loss and harm from either.
Machine learning changes the scale and shape of that risk. Complex models can find patterns traditional methods miss, but they can be harder to explain, more sensitive to shifts in data and more likely to encode historical bias. Generative AI introduces new failure modes, such as confident but false output, and new dependencies on a small number of providers. When many institutions use similar models, vendors or data, their errors can be correlated, so a single flaw or a single change in conditions can affect much of the market at once.
The consequences reach customers and the wider economy: unfair credit decisions, mispriced risk, inadequate reserves, procyclical behaviour in stress, and operational outages when a widely used service fails. Model governance is how the industry keeps those risks visible and controlled.
Grounded in method, linked to science and governance.
The platform applies artificial intelligence and machine learning, AI evaluation and assurance engineering, uncertainty quantification and model validation and probability and statistics from the STEM Commons hosted by The Global Centre for Risk and Innovation (GCRI). See the STEM foundation page.
At GCRI, Artificial Intelligence and Autonomous Systems studies AI capability and evaluation, and Future of Work and Skills its effect on work. At the Global Risks Forum (GRF), AI and Digital Governance, Innovation Nexus and Capital Nexus convene institutions on how AI and finance are governed. Within GRA, Financial Stability, Financial Regulation Nexus and Actuarial and Quantitative Risk are close neighbours. Work can draw on the Foundry systems NXS-NXQ (software, models and capability lifecycle), NXS-LAB (research, experiments and simulations) and NXS-NSF (trust, authorisation and assurance) and contribute to the Methods and Technical and Exponential Technology series. See counterparts across the Consortium.
The areas of artificial intelligence this platform takes on.
AI and Financial Model Risk studies how the models that banks, insurers and asset managers use to lend, price, reserve, trade and serve customers can be wrong or misused, and how machine learning and generative AI change that risk. Its pages set out model risk governance, credit and pricing models, generative AI, explainability and fairness, monitoring and drift, human oversight, third-party models and stress models, with every source labelled. Global Risks Alliance shares industry understanding and does not validate models or give advice.
Model risk governance
Inventories, tiering, validation, approval and the board and senior management oversight that model risk management requires.
Credit and pricing models
Scoring, probability of default, loss given default, insurance pricing and the data and assumptions they rest on.
Generative AI
Large language models in customer service, coding, research and operations, and the new failure modes they bring.
Explainability and fairness
Explaining model outputs to customers, staff and supervisors, and testing for unfair or unlawful outcomes.
Monitoring and drift
Tracking model performance after deployment and detecting when data or relationships shift.
Human oversight
Where people review, override or remain accountable for model-driven decisions, and how to keep oversight real.
Third-party models
Vendor scores, foundation models and data products that firms use without full access to their design.
Stress and scenario models
Models that project losses and capital under stress, and the special risks of models used for rare events.
Built on named science, technology, engineering and mathematics.
Every claim on AI and Financial Model Risk rests on methods with a canonical page in the STEM Commons, so readers can see what a finding depends on and where its limits lie.
How work moves in this field.
A question becomes evidence, a governed build, an organised community, a controlled test, a reviewed publication and, finally, a decision taken by the competent authority under its own rules.
- Question Universe National, regional or institutional priority. Question in AI and Financial Model Risk
- Evidence Foundry STEM foundation of the platform. Evidence in AI and Financial Model Risk
- Build Foundry Foundry. Build in AI and Financial Model Risk
- Organise Universe Nexus ecosystem. Organise in AI and Financial Model Risk
- Test Universe Universe (independent site). Test in AI and Financial Model Risk
- Publish Reports Reports. Publish in AI and Financial Model Risk
- Decide Competent authority Competent authorities and institutions, through NXS-RLS (participants, agreements and handoffs). Decide in AI and Financial Model Risk
- Renew Universe Corrections and the next annual agenda. Renew in AI and Financial Model Risk
Each step of the cycle in the pillar it runs in; the decision stays with the competent authority, outside the programme.
- Labs
- Foundry
- Universe
- Nexus Standards
- Reports
- Risk Academy
- Competent authority (outside the programme)
How this works across the six pillars.
AI and Financial Model Risk works through six pillars of one annual programme, led by the Nexus Consortium, with GCRI, GRF and GRA each contributing in its role: explore, experiment and prototype in Labs, build, simulate and prepare in Foundry, convene and test in Universe, codify, correct and apply in Nexus Standards, publish in every format in Reports, teach in the open in Risk Academy.
handoff in playrecorded handoffno direct handoff
Programme yearAll phases
Route
The whole programme
Every pillar hands its work to another; together they close the year’s feedback loops.
L · F · U · S · R · A
- Pillars
- 6 of 6
- Handoffs
- 18 of 18
- Loops
- 17 closed
Handoffs in play
- F01Universe → Foundry · Admitted mission becomes a cycle project with a decision caseReadiness preparation
- F02Foundry → Universe · Core Build Report and go or no-go evidenceControlled build
- F03Foundry → Reports · Sealed observations and indicator series behind a profileCloseout
- F04Universe → Reports · After-action and scenario findings for chapters 10 and 13After-action review
and 14 more
- LabsWhere questions become experiments and prototypes: new methods, data and tools are tried in the open, with their limits on record, before the Foundry builds them, under the Nexus Consortium.Labs
- FoundryWhere exposure and scenario evidence is built so supervisors, insurers and investors can read it.AI and Financial Model Risk in FoundryFoundry
- UniverseWhere scenarios are tested in controlled exercises, and finance-readiness dialogues read what they mean for capital and cover.AI and Financial Model Risk in Universeuniverse.nexus
- Nexus StandardsWhere disclosure and data rules are codified so risk evidence is comparable for finance and insurance.Digital, Cyber and AI in Standardsnexustandards.com
- ReportsWhere sector analysis and the finance and insurance readiness series are published, with exposure and limits stated.AI and Financial Model Risk in ReportsReports
- Risk AcademyWhere finance, insurance and public-finance professionals learn to read risk evidence, on Risk Academy.Faculty of Technology and Digital Systems in Academyrisksacademy.com
The same field, seen three ways.
GCRI studies a field as science, GRF governs it and GRA builds shared financial understanding of it. AI and Financial Model Risk links to its counterparts so that work on one side is visible on the others.
- Artificial Intelligence and Autonomous SystemsInstitutionGCRIShared fieldArtificial intelligence
- Future of Work and SkillsInstitutionGCRIShared fieldArtificial intelligence
- Innovation NexusInstitutionGRFShared fieldArtificial intelligence
- Capital NexusInstitutionGRFShared fieldFinancial stability
- AI and Digital GovernanceInstitutionGRFShared fieldArtificial intelligence
Pages in AI and Financial Model Risk.
Industry scope and publisher
PageFinancial discipline overview
PageProfessional research
PageTechnology and emerging practices
PageResilience and finance readiness
PageResearch and industry programmes
PageAnalytical and data resources
PageProfessional community
PageProfessional learning
PageProfessional meetings
PageReports and sector briefs
PageMembership and research participation
PageContact the industry platform
PageAntitrust, independence and financial perimeter
PageSTEM foundation of AI and Financial Model Risk
PageHow work moves in this field
PageBuild in Foundry
PageTest in Universe
PagePublish in Reports
PageThe same field across the Consortium
PageTopics
Where this page leads across the Consortium.
Questions and answers.
Who publishes AI and Financial Model Risk?
Is AI and Financial Model Risk active?
How can I contribute?
Where are findings published?
Which values on this platform are real?
How do I report an error?
Does the platform validate or approve models?
How is it different from general AI safety work?
Contribute evidence from this page.
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.
- EmailGateway GW-GRASend it from this pageEvidence, a correction, a case, a method, a review, a translation or profile input. It arrives tagged with this page, so the editor who owns it receives it.
- ZenodoOpen depositDeposit it and submit to the communityPublish your own record on Zenodo with your ORCID iD, then submit it to the Reports community there. Curators review it; the DOI stays yours.
- ReportsGlobal Risks AllianceSee every way to contributeThe contribution types, how each is reviewed, what credit you receive, the licence and what is never accepted.
Never send personal data, sensitive infrastructure detail or anything you have no right to share. A contribution is reviewed before use; sending it confers no authorship or editorial rights, and sponsorship never buys a finding.
How AI and Financial Model Risk works.
Choose a view and press Play to follow the platform’s process, its programme year, the standards it applies and how its findings are published. The outline shows how the work is designed to run, not the status of any decision.
AI and Financial Model Risk
How AI and Financial Model Risk works
Platforms · Programme year · Standards · Reports
A platform joins a field to named methods, labelled evidence and the outputs it publishes, without becoming the authority it serves.
01 Scope
- 01Field and familiesStart · The platform leadThe home states the field, its families and who it serves.
- 02Topic dossiersRecord · Platform authorsTopics explain the subject with labelled evidence.
02 Evidence
- 03Evidence labelsRecord · GCRIEach value is labelled: observed, published statistics, estimated, modelled, scenario or unknown.
- 04MethodsStep · Method ownersMethods come from the STEM Commons, each with one canonical page.
03 Work
- 05Scenarios on the systemsModel · Foundry pillarScenarios run on the Nexus systems, labelled as scenario.
- 06Standards that applyRecord · Nexus StandardsThe platform names the standards that apply. A validator report is evidence for review.
04 Publish
- 07EditionsRecord · ReportsFindings are published through Reports with their limits.
- 08Authority decidesOutcome · The competent authorityDecisions and execution stay with the government, operator or licensed institution.
AI and Financial Model Risk home · Platforms · Shared methodsAn outline of how the work is designed to run; it does not report the status of any decision or record.
Take part in AI and Financial Model Risk.
Start from who you are. Answer an open call. Or form and lead a competence cell on the AI and Financial Model Risk platform.
National institutions
Ministries, statistics offices and agencies: a national desk and chapter, and a seat in the working groups. National chapters
Sectors and professions
Operators, regulators, banks, insurers and professional bodies whose work touches AI and Financial Model Risk. Participate
Learning
Courses and practice, with a record of every course completed. Free to study. Open learning
Standards
Draft, review and adopt the methods and data standards with the working groups. Working groups
Communities and research
Universities, civil society and individual experts: projects, data and review. Write to the team
| Open call | Seeks | Closes |
|---|---|---|
| Founding call for experts to the working groups CALL-FOUNDING-EXPERTS | expert, editor, chair | 2026-12-15 |
| Open nomination for the three independent expert seats of the Global Standards Council CALL-GSC-IND | council-member | 2026-12-01 |
| Nominations of interim regional members (applies only if the transitional provision is decided) CALL-GSC-REG | council-member | 2026-12-15 |
| Review of the wave-1 risk core outlines (C1, C2, C3, C4, C8, C10, C11) CALL-REVIEW-WAVE1 | contributor | 2027-01-31 |
| Experts for the core standards of risk science: uncertainty and evidence grading, scales, structured expert judgement, tolerability, extremes and compound events, catastrophic and existential risk, ethics and equity, units and measurement (NRS-C17 to NRS-C24) CALL-CORE-SCIENCE | expert, editor, chair | 2027-01-31 |
| Independent reviewers for the families of the hazard atlas, three from at least two regions for each family (NHS2-R19), priority families first CALL-ATLAS-REVIEW | expert | 2027-03-31 |
| Members of the interim Frontier Safety Review Panel and red-team reviewers (biosafety and biosecurity, AI safety and security, critical-infrastructure security); applications treated in confidence CALL-FRONTIER-SAFETY-REVIEW | expert | 2026-11-30 |
| Experts in ontology, metrology, software verification and conformity assessment for the requirement register, the NHS-2 machine representation, the conformance test suites and the conformity assessment scheme CALL-CONFORMANCE-ONTOLOGY | expert, editor | 2027-01-31 |
Calls past their closing date are hidden automatically. Every call, proposal and route: open areas and calls.
Gather the cell
Bring together three to seven people with complementary skills: data, field practice, policy and communication.
Register it
Write to the team with the cell’s name, members, area and region. You receive a reference and a contact. Write to the team
Choose a charter
Pick one focus: research and evidence, data and maps, standards review, or community preparedness.
Learn together
Follow shared study paths at Risk Academy, with practice and a record of every course completed. Hazards and resilience
Lead and report
Hold regular sessions. Publish what the cell learns through Reports. Bring findings to the working groups. Working groups
What a national team or a local community can do on The Global Risks Index (NXS-GRX) today.
Your country’s profile
Open risk profiles for 195 countries, with sources and licences on every value. Open the Index
Correct or reply
Correct a value or reply for your country. The change is logged with its effect. Open the Index
Offer data
Offer a dataset. It enters through the licence gate, with your credit. Open the Index
Every figure on this page is indexed: hazards, climate, conflict, the world.
The Index is the Consortium’s indexing system for risk data. Every indicator, hazard, series and country profile carries its Nexus Standards identifier, its source and its licence, so what this page shows can be traced, corrected and reused.
- 01 · Sources205 registered sourcesEvery value keeps its source and licence; withheld values are counted.Data and licences
- 02 · Nexus StandardsNIS-1 and NHS-1451 indicators, 905 hazards; every record carries a UNI.Nexus Standards
- 03 · The Index54 series, 195 country profilesComputed on the standards; no value is published until an edition is released.The Global Risks Index
- 04 · This pageAI and Financial Model RiskMaps, figures and feeds read the Index in the context of hazards, climate, conflict.Public API
Series in this context hazards, climate, conflict
- GLBGlobal SystemsThe state of interdependent global risk: the twenty domain indices aggregated to the world, the five core systems (water, energy, food, health, biodiversity) and their couplings, the global tail register and the gap…NXS-GRX-AR-000002Hazards · Climate · Conflict
- THDisaster Risk Intelligence (DRI)Signals, observations, event streams, geospatial layers and scenario packs turned into structured risk intelligence for preparedness and planning.NXS-GRX-AR-000111Hazards · Climate · Conflict
- GSGlobal CitiesUrban exposure, heat, flooding, informal settlement, service dependency and the economic weight of cities at risk.NXS-GRX-AR-000068Hazards · Infrastructure · Climate
- GSGlobal ClimateChronic and acute physical climate hazard, transition exposure, adaptation gaps and scenario pathways to 2030, 2050 and 2100.NXS-GRX-AR-000065Climate · Hazards · Finance
Country profiles the world
Every country profile carries the source and licence of each value. Correct a value or reply
Hazard domains NHS-1
- Atmosphere and weatherNXS-GRX-AR-000135
- Solid EarthNXS-GRX-AR-000138
- Systemic and compound eventsNXS-GRX-AR-000148
Geodata and API open
Open layers on an equal-area hexagon grid, shown as densities per km² or as symbols: People · GHSL 2025People · WorldPop 2025Built-up areaPortsAirportsRoadsRailwaysRivers and lake centrelines
/api/v1/public/grix · https://nexustandards.com/uni/{uni}
Developers · Data · Made with Natural Earth, Public domain; GHSL and WorldPop, CC BY 4.0
Snapshot of the Index registries pinned 2026-10-23 · 22c6eed. Research and risk-analytics information: not a rating, an official classification, a warning, a benchmark or advice.
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.
Legal documents.
Maps, data and figures on this site are shown for context and general information. They are not advice, forecasts or warnings; the limits of the information are set out in the Disclaimers and Data Notices.