Skip to content
TrustList
News

MAS sets AI risk rules for all Singapore financial institutions

Editorial

By TrustList Editorial

The guidelines cover every kind of AI, including models supplied by third parties, and phase in over two years: core duties from October 2027, the rest from October 2028.

About MAS sets AI risk rules for all Singapore financial institutions

MAS sets AI risk rules for all Singapore financial institutions

7 October 2026: The Monetary Authority of Singapore has issued its Guidelines on Artificial Intelligence Risk Management for Financial Institutions, which apply to every financial institution in the country and to every form of AI it uses. The guidelines take effect on 7 October 2027, giving firms a year to prepare.

MAS describes the document as its supervisory expectations for responsible AI adoption. It covers banks, insurers, payment institutions and financial advisers alike, and lets each firm tailor its controls to its own risk profile and to the scale and nature of its AI use. A firm whose AI would not materially affect operations, customers or other stakeholders if it performed badly may keep its policies basic.

The expectations themselves are broad. Boards and senior management are made accountable for AI risk. Firms are expected to manage risk across the whole life of a system, including data governance, testing, human oversight, cybersecurity, monitoring and change management, and to keep an inventory of the AI they use. The Register reports that the guidelines also call for independent review of AI systems before they go into production, by people who were not involved in building them, and for fallback arrangements for high-risk uses.

For vendors, the sharpest line concerns third parties. The guidelines state that firms remain accountable for AI developed, operated or provided by third parties, and they must obtain adequate assurance from those providers. A supplier selling AI features into Singapore finance should expect questions about testing evidence, monitoring and the models underneath its product.

The timetable has two steps. The requirements in sections 3 and 4 are due by 7 October 2027, and those in sections 5 and 6 by 7 October 2028.

MAS Deputy Managing Director Ho Hern Shin said that realising the benefits of AI "requires financial institutions to understand and manage the risks."

Related on TrustList:

Sources

Categories & features

TrustList Weekly

The week in software and IT, in one email

The news that matters to buyers, new rankings and our own research. Every Thursday, free, and easy to leave.

We will email you to confirm. Unsubscribe with one click in any issue. Privacy policy