Financial institutions will remain accountable for use of third-party AI tools: MAS

By the Editorial Team · Thursday, October 8, 2026

The Monetary Authority of Singapore has issued new guidelines requiring financial institutions to maintain full responsibility for AI tools they use, including those from third parties, with staged implementation starting in 2027. These rules aim to ensure robust risk management, human oversight, and transparency to prevent biases and errors, supporting innovation while protecting customers and Singapore's financial stability.

Financial institutions in Singapore will remain fully accountable for the use of artificial intelligence (AI) tools, including those developed by third parties, under new guidelines issued by the Monetary Authority of Singapore (MAS). The updated regulatory framework, announced on October 7, sets out expectations for managing risks associated with AI applications in the financial sector, such as data breaches and inaccurate outputs.

The guidelines, which will be implemented in stages starting from October 2027, aim to address challenges posed by advanced AI models that operate with increasing autonomy. MAS emphasized that institutions—including banks and insurers—must assess factors such as model transparency, explainability, and contingency measures before deploying AI solutions sourced from external providers.

Institutions are required to maintain human oversight over AI systems and keep detailed inventories that document AI use cases, models, approved modes of operation, training data involved, and designated personnel responsible. Risk assessments must consider the potential impact of AI errors and the degree of system autonomy, with controls scaled proportionally. For high-impact applications such as credit decisioning and insurance underwriting, MAS calls for more stringent safeguards to prevent harms like bias or discriminatory outcomes.

The guidelines clarify that basic AI governance policies are adequate for lower-risk AI uses, such as drafting emails, summarizing internal meetings, or creating marketing content, provided they do not produce materially adverse effects on customers or the institution. Senior management and boards are expected to regularly review AI policies and risk exposures and establish clear internal channels to report incidents arising from AI use.

MAS also noted that existing governance structures, if sufficiently robust, may be adapted for AI oversight, and that financial institutions need not establish specialized AI committees solely for compliance purposes.

The new framework will be enforced in two phases: by October 7, 2027, institutions must have foundational governance measures in place, including AI inventories and risk assessment capabilities. By October 8, 2028, they are required to implement rigorous testing and continuous monitoring of AI systems, maintain strong human control, and secure necessary technical and staffing resources.

MAS plans to continue consultations in 2027 to refine guidance, especially concerning the use of autonomous agent AI.

Ho Hern Shin, MAS’s deputy managing director, highlighted AI’s potential to enhance customer experiences, strengthen risk management, and drive innovation within the financial sector. However, he stressed that realizing these benefits sustainably depends on robust risk management and regulatory clarity, enabling institutions to innovate confidently while maintaining customer trust and safeguarding Singapore’s financial stability.

This article was reported by synthesizing coverage from multiple independent news outlets to present a balanced, unbiased account. Read about our editorial standards. Spotted an error? Let us know.