Singapore tests how to deploy AI agents with safeguards
A minister told parliament that agencies are running trials while the government reviews how existing accountability and security rules apply to autonomous agents.
Trials inform a review of safeguards
Singapore’s government is running trials and experiments to learn how to deploy agentic AI responsibly, Senior Minister of State for Digital Development and Information Tan Kiat How told parliament on October 7, according to CNA. He said agencies including GovTech are testing systems with safeguards while the government reviews how its frameworks should adapt to agents that can act, interact with external systems and change behavior with less human oversight. The statement describes ongoing evaluation; it does not announce that a new regulatory framework has been adopted or that a particular agent is cleared for unsupervised use.
Existing duties still apply
Tan said existing guidelines and regulations, including risk management and oversight, apply to agentic systems. He emphasized human accountability, controls and cybersecurity for the systems that agents touch, and urged private organizations to exercise care even outside critical infrastructure. CNA also reported a separate written parliamentary reply from Digital Development and Information Minister Josephine Teo on October 6: Singapore is studying stronger safeguards for high-risk AI uses, and she said no cyberattack against government agencies involving an unsupervised AI agent had been reported so far. That is a statement about reported incidents, not proof that such attacks cannot occur. The government’s review and the results of its trials remain open.