Netherlands DNB AI: Principles 2019, Good Practices for Audit Trail & AFM Sustainable Finance AI
De Nederlandsche Bank (DNB) and the Autoriteit Financiële Markten (AFM) have developed a comprehensive AI governance framework for Dutch financial institutions. DNB's 2019 AI Principles were among the first detailed regulatory AI governance publications in Europe, establishing fairness, accountability, transparency, and ethics as the foundational framework. DNB's 2022 report on 'Good Practices for AI in Financial Sector' identified specific deficiencies at Dutch supervised institutions that have become the baseline for DNB examination of AI governance.
DNB Report on Good Practices for AI in Financial Sector — 2022
Published: 2022
Scope: All DNB-supervised credit institutions, insurance undertakings, and pension funds
Key findings: Majority of institutions lacked adequate AI governance frameworks; audit trails for AI decisions were insufficient for regulatory and legal purposes; bias testing was not conducted systematically; explainability capabilities were inadequate for consumer-facing AI decisions
DNB expectation: DNB expects supervised institutions to maintain comprehensive audit trails for all AI decisions that significantly affect consumers or capital requirements — absence of audit trail is itself a governance finding
FATE framework: DNB uses Fairness, Accountability, Transparency, and Ethics (FATE) as the evaluative framework for AI governance examinations
Source: DNB AI Governance — dnb.nl
Regulatory Risks and Compliance Challenges
The AFM has issued guidance on AI use in investment research, investment advisory services, and sustainable finance disclosure — identifying specific compliance requirements for AI systems used to generate ESG ratings, climate risk assessments, and sustainable investment classifications. As the EU Sustainable Finance Disclosure Regulation (SFDR) and EU Taxonomy Regulation create AI-assisted classification challenges, the AFM has flagged AI greenwashing risk as a priority enforcement concern.
The Netherlands' implementation of the EU AI Act adds to the existing DNB and AFM AI governance frameworks. Dutch institutions subject to both DNB/AFM examination and EU AI Act requirements must coordinate between existing national AI governance (FATE framework) and the EU Act's conformity assessment requirements. The Netherlands Authority for Digital Infrastructure (RDI) coordinates with DNB and AFM on EU AI Act implementation for financial sector AI.
Claire's AI Compliance Solution
Claire Platform Capabilities
DNB FATE Framework Compliance Documentation
Claire implements the DNB FATE evaluation framework — generating fairness testing results, accountability documentation, transparency evidence, and ethics review records that meet DNB examination standards for AI governance in supervised Dutch financial institutions.
AI Audit Trail for DNB Compliance
Claire's audit trail module maintains immutable, comprehensive records of all AI decisions with the timestamp, input data, model version, and output values that DNB examiners require — addressing the audit trail deficiency that DNB's 2022 Good Practices report identified as the most common AI governance failure.
AFM Sustainable Finance AI Compliance
Claire's ESG compliance module reviews AI-generated sustainable finance classifications and ESG ratings for compliance with SFDR and EU Taxonomy requirements — addressing AFM's AI greenwashing concerns by ensuring AI sustainability classifications are based on disclosed, verifiable criteria.
Compliance Checklist
AI Regulatory Compliance Requirements
AI governance framework with board oversight: Board-approved AI policy covering all AI systems with named accountability owners.
Pre-deployment risk assessment: Written risk assessment for all material AI systems before production deployment.
Independent model validation: Annual independent validation of AI models with documented results.
Fairness and anti-discrimination testing: AI models tested for disparate impact on protected groups before deployment and annually.
Explainability for affected individuals: AI decisions affecting consumers include explanation capability meeting applicable regulatory standards.
Third-party AI vendor oversight: Due diligence and ongoing oversight documentation for all AI vendor relationships.
Data quality and governance: Training data quality documented, lineage tracked, and reviewed for bias before use.
Consumer protection compliance review: AI customer-facing tools reviewed against applicable consumer protection laws.
Incident response for AI failures: Written incident response plan with regulator notification protocols for AI material failures.
Examination-ready documentation: All AI governance records maintained for regulatory access within 48 hours of request.
Frequently Asked Questions
What is DNB's FATE framework for AI governance?
DNB's FATE framework — Fairness, Accountability, Transparency, and Ethics — is the evaluative framework DNB uses to assess AI governance at supervised institutions. Fairness requires AI systems to produce outcomes without prohibited discrimination. Accountability requires clear human accountability for AI decisions and audit trails. Transparency requires explainability to affected individuals. Ethics requires AI to align with societal values and institutional ethical standards. DNB examiners assess all four dimensions in AI governance reviews.
What did DNB's 2022 Good Practices report find?
DNB's 2022 Good Practices report found that the majority of Dutch supervised institutions had significant AI governance gaps. Specific deficiencies included: absence of comprehensive AI audit trails (the most common finding); inadequate bias testing (most institutions had not systematically tested AI models for discriminatory outcomes); insufficient explainability capabilities for consumer-facing decisions; and weak accountability frameworks with unclear ownership of AI decisions. DNB has since made AI governance a standard examination component.
How does AFM supervise AI in sustainable finance?
AFM is actively monitoring AI use in ESG rating, climate risk assessment, and sustainable finance classification — areas where AI is increasingly used to generate SFDR and EU Taxonomy disclosures. AFM has identified 'AI greenwashing' — AI systems classifying products as sustainable based on criteria that do not actually meet the regulatory definition — as a priority enforcement concern. AFM expects AI-generated sustainability classifications to be based on verifiable, disclosed criteria that can withstand examination scrutiny.
How do Dutch institutions navigate EU AI Act and DNB requirements together?
Dutch financial institutions must satisfy both DNB's national AI governance requirements (based on the FATE framework and Good Practices report) and the EU AI Act's conformity assessment requirements for high-risk AI. In practice, these frameworks are compatible — both require bias testing, audit trails, explainability, and governance documentation. Institutions should design governance frameworks that satisfy both simultaneously, with DNB FATE documentation serving as evidence for EU AI Act technical documentation requirements.
What audit trail requirements does DNB apply to AI?
DNB expects supervised institutions to maintain audit trails that document: the version of the AI model used for each decision; the input data values that drove the specific decision; the model output and the decision based on that output; the timestamp of the decision; and the downstream action taken (e.g., credit approved/denied, alert escalated/closed). Audit trails must be maintained for regulatory retention periods and must be accessible to DNB examiners within a defined timeframe of their request.
Related: Finance AI Overview | AI Model Risk Management | Regulatory Compliance