Singapore’s financial regulator has set out staged expectations for how financial institutions should manage AI, including generative AI and increasingly autonomous agents. The framework asks firms to match oversight and controls to the risks of each use, rather than treating every AI tool alike.
Watch Desk analysis
What happened
The Monetary Authority of Singapore (MAS) published its Guidelines on Artificial Intelligence Risk Management on 7 October. Finance Magnates says they cover financial institutions and AI technologies broadly, from traditional machine-learning models to generative AI and AI agents.
The implementation is staged: firms are expected to establish AI oversight, risk management and policies from 7 October 2027, followed by broader lifecycle controls, capabilities and infrastructure from 7 October 2028. MAS also plans to consult the industry in 2027 on whether further guidance for agentic AI would be useful. Finance Magnates’ account of the guidelines describes them as supervisory expectations, not a standalone law imposing identical technical requirements on every firm.
Why it matters
The practical principle is proportionality: firms should understand where AI is used, assess how important and risky each application is, and apply controls accordingly. Proofreading an internal email is not the same proposition as influencing investment recommendations or customer risk assessments.
MAS specifically warns that agents with access to tools could take erroneous or unauthorised actions. The guidance also identifies financial, operational, conduct, financial-crime and reputational risks, alongside security, privacy and third-party risks, particularly for generative AI.
Our read
This is more than a broad instruction to “be responsible”. The staged dates give financial firms a timetable, and the risk-based approach gives them a way to distinguish low-stakes assistance from AI that can affect customers or financial decisions. The useful test will be whether firms can show what AI they use, what risks they identify and why their controls fit.
What to watch
- How firms prepare for the first expectations taking effect on 7 October 2027.
- Whether MAS’s planned 2027 consultation leads to specific guidance on agentic AI.
- How supervisors assess whether controls are proportionate to each use and its risks.
Discussion spark: Should financial regulators set common minimum controls for every AI system, or let firms scale safeguards to the risk of each use?
Sources and evidence
- Source update (12 October 2026, 00:00 UTC)
Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.