Singapore runs AI agent trials as it weighs tighter oversight rules
Source: CNA Tech
Singapore is trialling AI agents inside government to learn how autonomous systems can be deployed responsibly, with Tan Kiat How saying oversight frameworks are being reviewed.
Singapore's government is testing AI agents in live settings to work out how to roll the technology out responsibly, with Tan Kiat How saying that effort sits alongside a wider look at whether today's safeguards still do the job. Tan, the Senior Minister of State who oversees digital development and information, frames the trials as groundwork for rules that can cope with software capable of acting on its own.
That distinction matters because AI is changing character. Where earlier tools summarised or classified, newer systems can pursue a goal, string together steps and carry out work with minimal oversight. Officials are now asking whether the frameworks governing such deployments are pitched at the right level of autonomy.
The government's approach runs on two tracks. Trials serve as a practical laboratory, showing officials how agents behave once embedded in real processes. The parallel policy review takes those lessons and asks how protections should be sharpened so oversight keeps pace as autonomy grows.
Industry should read this as a cue to prepare rather than wait. Vendors and adopters of agentic software can expect harder questions about how much latitude a system gets and who answers for its actions. Building accountability into a deployment from the start is far easier than bolting it on afterwards.
Why it matters for Singapore: Clear, dependable rules are part of what makes the country an attractive base for technology firms. As agentic AI moves from novelty to everyday infrastructure, the frameworks now under review will shape what businesses here can build, what they can sell abroad and how far the public will trust automated decisions. Testing before legislating lets Singapore keep that credibility intact.
