Josephine Teo Calls for Careful Agentic AI Testing Before Wider Enterprise Use

Agentic AI — autonomous systems that act on behalf of humans without moment-by-moment oversight — is shaping up to be the next big test for Singapore’s AI governance framework. Speaking at IBM Think on Tour Singapore, Minister for Digital Development and Information Josephine Teo delivered a clear message: the tools businesses are rushing to deploy may outpace the safeguards designed to contain them.

Teo warned that existing risk frameworks were built for an earlier generation of AI where humans stayed in the decision-making loop. Agentic AI changes that calculus fundamentally. “When agents are designed to make decisions on behalf of humans, the kinds of errors, risks, and safety issues that arise are honestly not completely well understood,” she said, adding that it will take time before the full range of problems is appreciated.

Her remarks come as Singaporean enterprises across banking, logistics, healthcare, and government begin experimenting with autonomous AI agents — from customer service bots that independently escalate issues to procurement agents that negotiate with suppliers. The gap between deployment speed and governance readiness is not unique to Singapore, but the city-state’s role as a regional AI hub means its regulatory approach will be closely watched by neighbouring markets.

Teo outlined a practical path forward: clearly define each agent’s purpose, design controls appropriate to the level of autonomy, and build containment measures before problems surface. Singapore plans to collaborate with technology partners to develop better testing methodologies drawn from real-world deployments, rather than relying on theoretical risk models alone.

Why it matters for Singapore: The government has placed trust and governance at the centre of its AI strategy, and Teo’s speech signals that agentic AI testing frameworks are a near-term priority. For businesses deploying autonomous systems here, the message is to build accountability into agent design now — before regulation catches up. Singapore’s approach to agentic AI governance could well set a template for the region, mirroring how its Model AI Governance Framework influenced policy across Southeast Asia.