Posted in

Singapore Explores Using Web Domain System To Track AI Agents

AI Demand
Singapore Export Outlook Supported by AI Demand Despite Risks

Singapore’s Infocomm Media Development Authority (IMDA) is exploring whether the internet’s domain name system (DNS) can be adapted to identify and verify AI agents operating across platforms and organisations, Biometric Update reported, citing The Straits Times.

According to the report, the government is assessing whether existing internet infrastructure can track autonomous AI agents, confirm who controls them and establish accountability. It is also considering whether a dedicated national registry for AI agents is needed.

Industry demonstration

The report said GoDaddy, Infoblox, Cloudflare and Google presented a “Chain of Trust” architecture in October, developed with IMDA, to verify the identities of AI agents during transactions across independent platforms.

Senior Minister of State for Digital Development and Information Jasmin Lau was quoted as saying the government would work with industry to find out what works in practice and build on existing infrastructure and emerging international standards. She added that it would step in with common standards or infrastructure where there are gaps industry cannot fill alone.

“Agents can act for us, but they cannot answer for us,” she said, according to the report.

Member of Parliament Tin Pei Ling has proposed a national registry modelled on Singpass that would link AI agents to accountable organisations and their authorised activities, the report said. In Parliament, she suggested a form of “biometric passporting” for AI agents, built on open and interoperable standards.

Government AI infrastructure

Separately, the Government Technology Agency of Singapore (GovTech) is developing GovHarness, shared infrastructure for building and operating AI agents across the government.

Speaking at GovTech’s STACK Conference on 5 November, Senior Minister of State Tan Kiat How said GovHarness will support deployment, connections to tools, and monitoring and evaluation, with common guardrails and evaluations.

He said GovTech has turned principles from IMDA’s Model AI Governance Framework for Agentic AI into technical requirements for government through its Agentic Risk and Capability (ARC) Framework. Under it, higher-risk actions such as financial transactions, official communications and data deletion need stronger controls, including human approval.

Tan also said GovTech’s Rabbit Deploy platform has more than 8,000 users across 90 agencies, with over 5,000 active applications.

No timeline has been given for a decision on a national AI agent registry.

Leave a Reply