GovTech Builds AI Agent Registry for 150,000 Singapore Public Officers

Singapore is about to put AI agents in the hands of 150,000 public officers — and it's building the accountability infrastructure first. GovTech Singapore is developing a registry of AI agents as part of its AI Assistant Desk suite, designed to give visibility over how civil servants use AI in daily tasks.

The registry tracks who owns each agent and what it does, providing a layer of oversight that Chief Executive Goh Wei Boon says is essential for secure deployment. "We want to have a layer of customisable rules, sanctioned AI tools and a registry to provide better visibility and security," he explained.

The security framework includes guardrails that prevent agents from deleting files or emailing external recipients, caps the number of email recipients to prevent spam, and runs automated checks to catch offensive language before it enters or leaves the system. These controls work regardless of whether agencies use in-house or third-party AI tools.

Unlike conventional chatbots, AI agents can proactively work out the steps needed to complete a task with minimal human input and respond to natural language prompts requiring no technical expertise. GovTech is currently piloting the suite with some public officers, with broader deployment planned for later in 2026.

The philosophy is straightforward: security is being baked into centralised platforms from the start, not treated as an afterthought. This approach reflects lessons learned from earlier government tech deployments where security considerations came too late in the process.

Why it matters for Singapore: At 150,000 users, this would be one of the largest government AI deployments in the world. The registry approach — tracking every agent's owner and function — could become a model for other public sectors wrestling with how to scale AI safely across large workforces.