The Monetary Authority of Singapore (MAS) has established its regulatory framework for artificial intelligence in finance, creating a structure that promotes innovation while managing risks effectively. The guidelines will become effective on October 7, 2027, with certain provisions rolling out by October 7, 2028. Financial institutions must now document their use of AI, evaluate associated risks, and implement controls that match those risks—whether the technology is developed internally or sourced from external providers.
A central aspect of the rules is proportionality, requiring firms to maintain supervision but allowing adjustments based on risk levels. Low-risk AI applications need only basic oversight, while high-risk systems demand strict governance. Ho Hern Shin, MAS Deputy Managing Director, said the aim is for firms to “innovate with confidence, while maintaining the trust of customers.” The framework also addresses AI supplied by vendors, obligating institutions to verify its appropriateness and apply extra safeguards if risks exceed acceptable thresholds.
Unlike other regulatory models, MAS does not require dedicated AI oversight committees. Instead, existing leadership teams can supervise compliance, provided they uphold strong controls throughout the AI lifecycle—from data management to ongoing monitoring and system updates. Periodic reviews ensure accountability persists as AI systems develop.
Related Post: Tech founders bet AI agent can cover Alaska news
Singapore’s strategy distinguishes itself by prioritizing material risks over technical specifications. Rather than imposing uniform rules for every AI application, it enables institutions to assess risks dynamically. This adaptability is essential in finance, where AI models and their applications evolve rapidly. The guidelines also include provisions for future discussions on agentic AI, systems capable of independent action, demonstrating MAS’s proactive approach.
Alex Goryachev, Innovator-in-Residence at Tulane University, described it as a framework others might adopt, arguing that risk-based regulation remains practical as AI progresses. He noted that vendor responsibility becomes manageable if institutions share assurance data instead of conducting separate audits. For agentic AI, he recommended defining clear ownership, setting permissions, and establishing rapid shutdown protocols, a gap MAS is addressing early in its planning.
Michael Heinrich, CEO of 0G Labs, supported the approach, stating that its core principles, proportionate supervision, institutional responsibility, and technical transparency, hold broad relevance. He emphasized the need for global standards to support cross-border AI operations, particularly in developing regions where AI could reduce costs for financial services lacking robust infrastructure.
Read Also: 32 African Countries Tax Foreign Digital Services
Nikki Stone, founder of YQ Advisory, commended MAS’s focus on material risks but stressed that accountability must extend to vendors. Even when third-party AI drives decisions, the financial institution remains liable for outcomes, she said. This aligns with MAS’s requirement for human oversight in all critical AI systems, regardless of their development source.
The staggered implementation, with some rules taking effect in 2028, gives firms time to prepare. MAS’s plan to consult on agentic AI next year shows recognition that current guidelines may need updates as AI capabilities grow. The timing also coincides with the Singapore FinTech Festival, where industry participants will discuss practical challenges in applying the rules.
Timing and Broader Implications for Global Financial Markets
The framework’s introduction precedes the Singapore FinTech Festival, an annual event scheduled for November 18–20 at Singapore EXPO. The gathering serves as a platform for the city’s financial technology sector to collaborate, potentially influencing how the new AI guidelines are adopted and refined.
