Loading
The Monetary Authority of Singapore published its Principles to Promote Fairness, Ethics, Accountability and Transparency (FEAT) in the Use of Artificial Intelligence and Data Analytics in November 2018, one of the first AI-specific governance frameworks issued by a major financial regulator anywhere. Formally voluntary, FEAT operates in practice as the standard MAS supervision assesses AI-using financial institutions against.
MAS FEAT Principles, the Monetary Authority of Singapore's 2018 principles for responsible AI in financial services, covering Fairness, Ethics, Accountability, and Transparency.
FEAT remains the primary AI governance framework for Singapore financial institutions a decade after publication. The MAS has supplemented it with the Veritas consortium tools (methodology for assessing FEAT compliance) and the AI Verify framework. FEAT is non-binding but supervisory expectations signal that large financial institutions are expected to demonstrate FEAT alignment. The four FEAT principles map reasonably well to later frameworks, NIST AI RMF Govern covers accountability, the EU AI Act covers transparency and fairness.
Source: MAS Principles to Promote Fairness, Ethics, Accountability and Transparency (FEAT) (2018)
AI-driven decisions do not disadvantage individuals or groups unintentionally. Goes beyond non-discrimination: institutions must also check that models do not perpetuate historical bias embedded in training data, even where that bias does not reference a protected characteristic.
AI use aligns with the institution's ethical standards, not just what the law technically permits. Ethical use is treated as a distinct expectation from legal compliance.
Internal governance, accountability structures and oversight remain robust as AI is adopted. Human decision-makers stay answerable for AI-assisted or AI-driven outcomes, an AI system is not a basis to disclaim responsibility.
Customers are given clear, simple explanations of what data is used and how, and are told when AI materially drives a decision that affects them, in language they can act on, not just a score.
FEAT is not a statute. MAS has not made it legally binding, and it carries no fixed penalty regime. In practice, supervisory conversations, thematic reviews and industry engagement consistently reference FEAT as the standard financial institutions' AI governance is assessed against, so non-adoption carries real supervisory risk even without a formal breach.
That is changing. On 13 November 2025 MAS issued a consultation paper proposing binding Guidelines on AI Risk Management, applying across the financial sector including generative AI and AI agents, with comments due by 31 January 2026. As of mid-2026 MAS has said it is reviewing consultation responses; the guidelines are not yet finalised. Separately, MAS-led industry effort continued through Project MindForge, whose second phase concluded in March 2026 with a jointly developed AI Risk Management Toolkit built with 24 banks, insurers and capital-markets firms.
FEAT states principles; it does not by itself say how to measure them. The Veritas Consortium, a MAS-led industry group, closes that gap with assessment methodologies published as a series of white papers, an enhanced Fairness Assessment Methodology, an Ethics and Accountability Assessment Methodology, and a Transparency Assessment Methodology, plus an open-source toolkit. They specify which fairness metrics fit which decision type (credit, insurance, investment advice), what thresholds should trigger investigation, and how to weigh competing fairness definitions against each other. Using the Veritas methodology gives an institution documented, defensible evidence of fairness assessment that satisfies MAS supervisory expectations.
What does MAS FEAT stand for?
FEAT stands for Fairness, Ethics, Accountability and Transparency, the Monetary Authority of Singapore's principles for the responsible use of AI and data analytics in the financial sector.
Is MAS FEAT legally binding?
No. FEAT is a set of supervisory expectations, not a statute. MAS has not made it legally binding, but supervisory conversations, thematic reviews and industry engagement consistently reference FEAT as the standard financial institutions are assessed against, so in practice adoption is expected.
Is that about to change?
It is heading that way. MAS issued a consultation paper on 13 November 2025 proposing binding Guidelines on AI Risk Management, with comments due by 31 January 2026. As of mid-2026, MAS has stated it is reviewing consultation responses; the guidelines have not yet been finalised. Institutions that have not implemented FEAT should not wait for the final rules.
What is the Veritas Consortium?
A MAS-led industry consortium (initially 27 financial institutions and technology firms) that publishes assessment methodologies translating FEAT into measurable criteria. Its white papers, including a dedicated Fairness Assessment Methodology, cover which fairness metrics fit which decision types (credit, insurance, investment advice) and what thresholds should trigger further investigation.
Does FEAT apply outside Singapore?
FEAT is a Singapore regulatory instrument, but because Singapore is a regional financial hub, institutions operating in or through Singapore across Southeast Asia are increasingly expected to demonstrate FEAT alignment, and FEAT has influenced AI governance frameworks elsewhere in ASEAN.
How does FEAT relate to NIST AI RMF and the EU AI Act?
FEAT predates both by several years (2018) and is sector-specific to Singapore financial services rather than a general-purpose framework. Its four pillars map loosely onto later frameworks: Accountability corresponds to NIST's Govern function, and Fairness and Transparency correspond to EU AI Act themes, but FEAT does not require conformity assessment or carry statutory penalties the way the EU AI Act does.
Last reviewed July 2026