Artificial Intelligence Model Risk Management: Observations from a Thematic Review
ARTIFICIAL INTELLIGENCE MODEL RISK MANAGEMENT OBSERVATIONS FROM A THEMATIC REVIEW
This information paper by the Monetary Authority of Singapore sets out good practices relating to artificial intelligence and generative AI model risk management observed during a thematic review of selected banks. It covers governance and oversight, risk management systems and processes, inventory and materiality assessments, as well as development, validation, deployment, monitoring, and change management standards. The paper aims to help financial institutions strengthen their responsible use of AI while balancing innovation and risk management.
Summary and content fields are drawn from the document text and are pending editorial review.
Who issued it
Who it governs
Financial institutions, risk managers, and compliance professionals should read this to understand supervisory expectations and adopt recommended good practices for AI model risk management.
What the document contains
The document addresses artificial intelligence and generative AI model risk management within the banking sector, outlining good practices across governance structures, AI identification and inventory management, risk materiality assessments, development and validation standards, deployment controls, monitoring, and change management.
It does not provide binding statutory rules or legal penalties, nor does it cover non-AI specific operational areas outside of model risk management.
- Financial institutions should establish robust model risk management frameworks tailored specifically to artificial intelligence and generative AI systems.
- Cross-functional oversight forums help coordinate AI governance and address evolving risks across different bank functions.
- Comprehensive AI inventories and risk materiality assessments allow institutions to apply commensurate controls proportionately.
- Independent validation and rigorous pre-deployment checks are critical to verify that AI models perform as intended before entering production environments.
- Ongoing monitoring for data and model drifts, combined with structured change management, ensures sustained operational safety for deployed AI models.
· 1. Overview · 2. Background · 3. Objectives and Key Focus Areas · 4. Governance and Oversight · 5. Key Risk Management Systems and Processes · 6 Development and Deployment · 7 Other Key Areas · 8 Conclusion
How it scores
Each score is backed by a quote verified against the document text. Editorial review has not yet confirmed them.
Human Dignity
0/5Respect for the inherent worth of persons
screening score · 2026-10-23
Human Oversight
4/5Humans remain in control of consequential decisions
Due to the uncertainties associated with Generative AI, banks continue to require human oversight or have a human-in-the-loop when using Generative AI to aid in decision-making.
7.1 Generative AI
screening score · 2026-10-23
Privacy
3/5Protection of personal data and private life
Output filters may be used to detect biasness or toxic language in the outputs of Generative AI and trigger a review by a human or another Generative AI model, or redact PII information in the outputs of Generative AI before they are presented to the user.
7.1 Generative AI
screening score · 2026-10-23
Impact Assessment
4/5Required assessment of an AI system's risks/impacts before or during use
In assessing risk materiality, most banks considered both quantitative and qualitative risk dimensions that could generally be grouped into three broad categories: a. Impact on the bank
5.3 Risk Materiality Assessment
screening score · 2026-10-23