Principles to Promote Fairness, Ethics, Accountability and Transparency (FEAT) in the Use of Artificial Intelligence and Data Analytics in Singapore's Financial Sector

Applies To

Overview

On 12 November 2018, the Monetary Authority of Singapore (MAS) published the FEAT Principles as a governance framework for Singapore financial institutions (FIs) using artificial intelligence and data analytics (AIDA). The FEAT Principles were developed in collaboration with the financial industry and are designed to be technology-neutral (not tied to any particular technology). They apply across a wide range of AI and advanced analytics use cases, from credit scoring and fraud detection to robo-advisory (automated investment advice) and insurance underwriting. Although the FEAT Principles are not binding regulations, they carry significant supervisory weight. MAS has indicated that it expects FIs to use FEAT as a reference when designing internal AIDA governance frameworks, and adherence is considered in supervisory engagement. To support practical implementation, MAS began working with the Veritas Consortium of financial institutions and technology companies in November 2019. Its task was to develop methodology documents and open-source tools (freely available software) that translate the FEAT Principles into measurable assessment criteria. Veritas first published a fairness methodology for credit risk scoring and customer marketing in January 2021. In February 2022 it covered all four principles, with insurance use cases in predictive underwriting and fraud detection. The FEAT framework is structured around four named principles. Fairness requires that individuals or groups are not systematically disadvantaged by AIDA-driven decisions unless the decisions can be justified, and that fairness considerations are embedded throughout the model lifecycle (from design through use and retirement). Ethics requires that AIDA use aligns with the ethical values of the institution and broader societal expectations, with human oversight maintained over consequential decisions. Accountability requires clear assignment of responsibility for AIDA outcomes within the institution, with AIDA use approved by an appropriate internal authority and the board kept aware, with appropriate model risk management and audit trails. Transparency requires that institutions maintain adequate internal documentation of AIDA systems and, where appropriate, provide customers with meaningful explanations of AIDA-driven decisions affecting them. MAS has integrated FEAT into its broader digital and innovation governance expectations and references it in the context of the Financial Services Industry Transformation Map.

Key Requirements

• •Establish an internal governance framework for AIDA that addresses the FEAT Principles, with AIDA use approved by an appropriate internal authority and the board kept aware.
• •Assess AIDA systems for potential discriminatory outcomes across protected characteristics; implement remediation measures where bias is detected.
• •Maintain comprehensive model documentation covering data sources, model design choices, validation results, and known limitations for all material AIDA applications.
• •Define and document human oversight mechanisms for AIDA systems making or informing consequential decisions about customers.
• •Provide customers with meaningful explanations of decisions materially influenced by AIDA where the decision adversely affects the customer.
• •Conduct model risk management activities proportionate to the materiality and risk profile of each AIDA application, including pre-deployment validation and ongoing monitoring.
• •Use the Veritas assessment methodologies to apply the principles, for example in credit scoring, customer marketing, insurance underwriting, and fraud detection.
• •Remain accountable for externally sourced AIDA models as well as internally developed ones.
• •Maintain audit trails sufficient to enable retrospective review of AIDA-driven decisions.
• •Periodically review and update AIDA governance frameworks to reflect changes in technology, business context, and regulatory expectations.

What Your Organization Must Do

• →Map which AI and data analytics uses in your firm touch customers, credit, underwriting or fraud detection.
• →Build or update an internal governance framework that addresses fairness, ethics, accountability and transparency.
• →Record who approved each AI use internally, and keep management and the board informed.
• →Document data sources, design choices, validation results and known limits for material AI applications.
• →Consider the Veritas methodologies and free open-source tools to turn the principles into testable criteria.
• →Confirm your firm stays accountable for AI models bought from vendors, not just those built in house.

Playbook Guidance

Step-by-step implementation guidance for compliance teams.

Governance Controls

Operational controls that implement requirements from this regulation.

Frequently Asked Questions

Are the MAS FEAT Principles legally binding on Singapore financial institutions?

No, the FEAT Principles are voluntary guidelines, not enforceable regulations. However, MAS treats them as carrying significant supervisory weight and expects regulated financial institutions to reference FEAT when designing internal AI and data analytics governance frameworks. Non-alignment may be raised during supervisory engagement.

Which financial institutions in Singapore are expected to follow the FEAT Principles?

FEAT is addressed to firms offering financial products and services in Singapore. Technology vendors are not in scope themselves; firms remain accountable for AIDA models they source from vendors.

What is the Veritas Consortium and how does it relate to FEAT compliance?

Veritas is an MAS-led industry consortium that turns the FEAT Principles into assessment methodologies and open-source tools. Its use cases cover credit risk scoring and customer marketing in banking, and underwriting and fraud detection in insurance.

Does FEAT require financial institutions to govern third-party and vendor AI systems?

FEAT says firms are accountable for both internally developed and externally sourced AIDA models. It does not prescribe specific vendor controls.

What level of seniority must own AIDA governance accountability under the FEAT Principles?

FEAT does not name an officeholder. It says AIDA use should be approved by an appropriate internal authority, and firms should keep management and the board aware of their AIDA use.

How do the MAS FEAT Principles compare to the EU AI Act for financial sector AI governance?

FEAT is a principles-based voluntary guideline focused exclusively on Singapore's financial sector, with no direct penalties for non-compliance. The EU AI Act is binding legislation with fines up to 35 million euros or 7% of global turnover, applies across sectors, and uses a risk-tier classification model rather than FEAT's four-principle structure.