First, MAS is positioning Singapore as a hub for advanced AI capabilities, Chia says. Over thirty financial institutions have already set up AI competency centres here, developing solutions that support both the local market and their global operations. These initiatives cover functions across the front, middle, and back offices. MAS encourages more financial institutions to establish their AI expertise in Singapore.
“To help financial institutions with advanced AI capabilities tackle more complex challenges, I am pleased to announce a new initiative — BuildFin.ai. This initiative will bring together technology providers and research institutions to collaborate with financial institutions on addressing shared, complex problems. The goal is to develop common resources and solutions that can benefit the entire financial ecosystem,” says Chia.
MAS and its financial institution partners have identified the first common challenge to address. Interestingly, Singlish poses a degree of linguistic complexity that current large language models are not yet fully equipped to handle. To tackle this, A*STAR will collaborate with financial institutions to develop a Voice-to-Text AI model tailored for the financial sector — one capable of transcribing conversations that feature Singlish as well as Singapore’s commonly spoken languages and dialects. Through this collaboration, partners can pool data, build a more effective model, and ultimately enhance customer service.
Second, MAS is supporting financial institutions that are at earlier stages of their AI adoption journey by providing shared resources to help them get started and scale effectively. PathFin.ai is a collaborative initiative developed with the industry, where financial institutions and technology firms share their AI adoption experiences and successful use cases. The platform features a curated library of industry-validated solutions and best practices, enabling institutions to save time and effort when identifying and implementing effective AI applications.
See also: Singapore Fintech Festival’s 10th edition drew over 70,000 participants from over 142 countries
Examples featured on PathFin.ai include an AI tool for optimising multi-currency cash management for corporate treasuries, and an agentic AI solution that automates end-to-end insurance claims processing. Since the platform’s launch in July this year, participation has expanded from 20 financial institutions to over 100 — and MAS continues to welcome more partners to join this growing initiative.
Third, on the discussion of governance and safety. As the financial sector increasingly adopts AI, strong governance and safety measures become crucial. In fact, the speed of AI adoption and the level of autonomy granted to AI systems largely depend on how robust the guardrails and controls are throughout the AI life cycle. Financial institutions have also expressed a desire for greater regulatory clarity.
“How has MAS responded? We chose not to outpace innovation with rigid, prescriptive regulations. Over two years ago, we launched an industry consortium called Project MindForge to build a shared understanding of AI risks and governance with the industry. The initial focus was on co-developing an AI risk framework to ensure a common comprehension of the main risks associated with AI usage, which was published in early 2024,” says Chia.
See also: JPM launches USD onchain deposit token, which acts like a bank deposit
In July this year, MAS was working on a set of AI supervisory guidelines. Today, MAS is releasing the Guidelines on AI Risk Management for consultation. These proposed guidelines outline expectations for financial institutions to identify AI risks and implement controls throughout the entire AI life cycle, scaled appropriately to the level of AI use and associated risks. They are principles-based rather than prescriptive, providing flexible guardrails to support responsible innovation in a rapidly evolving landscape.
At the same time, the Project MindForge consortium is also releasing an AI Risk Management Executive Handbook today. This handbook outlines the essential elements of effective AI risk management. Next year, it will be followed by a more detailed document with actionable insights and industry best practices, serving as a companion guide for financial institutions implementing the MAS Guidelines.
