Singapore Central Bank Issues New AI Risk Guidelines for Banks

Singapore’s central bank has issued new guidelines requiring financial institutions to identify, assess, and manage risks from their use of artificial intelligence, including third-party systems.

The Monetary Authority of Singapore published its first Guidelines on Artificial Intelligence Risk Management on October 7. They apply across the financial sector and cover all forms of AI, with the level of controls expected to depend on the scale of each institution’s AI use and the risks involved. 

Under the guidelines, financial institutions should maintain inventories of their AI systems, assess individual use cases, and apply controls throughout the AI lifecycle. These include data governance, testing, human oversight, cybersecurity, and ongoing monitoring. Boards and senior management are expected to establish clear responsibilities and risk management frameworks, although firms do not have to create dedicated AI committees if existing structures provide adequate oversight. 

The rules also address third-party AI. Institutions remain accountable for AI used in the services they provide, even when an external company develops or operates the technology. They should assess whether a provider’s systems suit their intended use and obtain sufficient assurances about the associated risks. If those risks cannot be managed within the institution’s risk appetite, it should consider restricting or suspending the service, or replacing it. 

The guidelines take effect on October 7, 2027, with phased implementation extending to October 2028. MAS also plans to consult the financial sector in 2027 on additional guidance for agentic AI systems, which can operate autonomously and access tools.