A backbencher proposed state infrastructure for agent identity; the government’s answer says trust needs three things answered — who the agent is, what it may do, and who is accountable.
What happened
On 4 November 2026, Singapore’s parliament debated an Adjournment Motion tabled by Tin Pei Ling, MP for Marine Parade–Braddell Heights, proposing a national digital identity infrastructure for AI agents — including a registry that other parties could query to check an agent’s identity and mandate. Jasmin Lau, Senior Minister of State for Digital Development and Information, replied for the government.
According to the published report of the sitting, Lau said confidence in agents depends on three questions being answered: who the agent is, what it is authorised to do, and who is accountable for its actions. She described work already under way rather than a new scheme:
- The Monetary Authority of Singapore is developing the Safeguards for Agentic Finance at Runtime (SAFR) with financial institutions and AI firms, to verify an agent’s identity and authority before a transaction executes while keeping auditable records.
- IMDA is working with GoDaddy, Infoblox, Cloudflare and Google on whether the internet’s Domain Name System can be adapted to identify and verify agents across platforms.
- GovTech is building an internal registry so agencies can track the agents they deploy and the officers responsible for them.
On the motion itself, the government’s position was that it supports the objectives and will work with industry and emerging standards, then “assess whether additional common infrastructure, such as a national registry, may eventually be needed”. Lau also noted that China, the United States and Estonia are exploring trusted agent identities, and that Singapore intends to use its 2027 ASEAN chairmanship to push common testing tools and standards for agentic AI.
Her closing line is the one worth keeping: “Agents can act for us, but they cannot answer for us. Our task is to make sure that behind every agent, there is always someone who does.”
Why it matters for KYA
A registry is a specific architectural choice, and a state is now weighing it. Registry-based KYA — a queryable, authoritative mapping from agent to principal — is attractive because a relying party gets one call instead of a chain of credentials to validate. It is also the choice with the most political weight: whoever operates the registry decides what an agent identity contains, who may issue one, and what happens when a record is stale. Until now that debate lived in vendor roadmaps; it is now in a national legislature.
The three questions are a usable acceptance test. Identity, authority, accountability. Any KYA scheme that cannot answer all three about a given agent at a given moment is incomplete, regardless of how good its cryptography is. That framing is more useful to a buyer than a feature list.
Note the sequencing: standards first, infrastructure later. The government’s stated approach is to build on emerging standards and assess common infrastructure after, not to stand up a registry first. For anyone planning agent deployments in Singapore, the practical reading is that DNS-based identification and runtime authorisation checks are the near-term interfaces, and a registry is a possible second layer.
It separates public-sector from market infrastructure. GovTech’s internal registry is an asset inventory — which agents exist, who owns them — not a trust service for third parties. The gap between “we know what we deployed” and “others can verify what we deployed” is exactly where KYA programmes fail, and the two are being built by different bodies here.
What to watch
Whether the assessment produces a consultation paper, and whether SAFR moves from a published framework — MAS released it on 3 July 2026 under its BuildFin.ai initiative, with reference code available on request — to something institutions must implement. Also watch the DNS experiment’s output: it is the only piece of this that could become cross-border infrastructure rather than national plumbing.
Sources
- AsiaOne, “‘Who answers for AI agents?’ Singapore working on safeguards as adoption grows, says Jasmin Lau” (report of the 4 November 2026 sitting), 4 November 2026 — asiaone.com
- The Straits Times, “Singapore explores adapting web domain name system to track AI agents, assesses need for national registry”, 4 November 2026 — straitstimes.com
- Monetary Authority of Singapore, “Safeguards for Agentic Finance at Runtime (SAFR)”, 3 July 2026 — mas.gov.sg
- Ministry of Digital Development and Information, “MDDI’s Response to PQ on Establishing Singapore as Global Leader in Agentic AI Security and Assurance”, 4 November 2026 — mddi.gov.sg
Know Your Agent (KYA) explains agent identity, verification and accountability. This is an explainer, not legal or compliance advice — see our Sources & methodology. New to KYA? Start with What is Know Your Agent? and the glossary.