Project Noor is an initiative of the BIS Innovation Hub that seeks to equip financial supervisors with independent, practical tools to evaluate and interpret the inner workings of artificial intelligence (AI) models used by banks and other financial institutions.
By combining explainable AI methods with risk analytics, the project aims to deliver a prototype through which supervisors can verify model transparency, assess fairness, and test robustness.
Why Noor
AI models now help approve mortgages, set card limits, and flag potential fraud in real time. While these services appear seamless, understanding why a model said yes, no, or "flag for review" can still feel opaque. Clear, human-readable explanations can strengthen confidence and help keep digital finance fair for everyone.
New regulations demand that high-risk financial AI be explainable and auditable. But there is no common, practical playbook for supervisors.
What is Noor
Led by the BIS Innovation Hub Hong Kong Centre in collaboration with the Hong Kong Monetary Authority (HKMA), the Financial Conduct Authority of the United Kingdom (FCA), and the Saudi Central Bank (SAMA), Project Noor ("light" in Arabic) will prototype the latest Explainable AI (XAI) techniques in a controlled setting. XAI converts complex model logic into plain language and intuitive visuals, making it easier to see which factors influenced a decision and how sensitive that decision is to change, all while preserving privacy.
What this prototype could mean in everyday terms:
- Greater transparency: Customers receive clearer reasons for credit decisions or fraud alerts.
- Consistent protection: Supervisors gain modern tools to check that similar customers are treated consistently.
- Responsible innovation: Banks can adopt new technologies with practical, privacy-preserving explainability checks.
It is important to note that financial institutions retain responsibility for model explainability and that Noor does not aim to prescribe definitive standards or replace existing practices. Instead, Noor strives to equip supervisors with methods and benchmarks to form their own informed opinions.