Singapore Is Not Legislating AI. It Is Exporting The Playbook Every Asian Regulator Wants

Singapore Is Not Legislating AI. It Is Exporting The Playbook Every Asian Regulator Wants

Singapore has never passed a comprehensive AI law, and at this point it probably will not need to. IMDA, the country's infocomm regulator, and the AI Verify Foundation have quietly built a governance stack that other APAC regulators are picking up wholesale. The most consequential AI policy decision in Asia right now is not a statute. It is a toolkit.

The Toolkit Is Now The Standard

The AI Verify Toolkit (AIVT), launched in 2022 and now at version 2.0.0a2, is the centrepiece. It is an open-source testing framework that scores AI systems against 11 internationally aligned principles, including fairness, transparency, robustness, and explainability^✦. In January 2026, IMDA added the world's first Model AI Governance^✦ Framework for Agentic^✦ AI, launched at Davos, to cover autonomous agent deployment. The newest AIVT release maps directly to ISO/IEC 42001:2023, the emerging international standard for AI management systems.

The export is already happening. Malaysia's MyDIGITAL initiative has cited AI Verify as a reference for its responsible AI^✦ rollout. Indonesia's Kominfo is piloting AI Verify-compatible testing for its banking AI guideline issued by OJK. Thailand's ETDA has folded parts of the framework into its voluntary AI ethics assessment. Three ASEAN regulators are effectively running Singapore's software as their national compliance benchmark^✦, without Singapore having passed a single AI statute.

Why The Soft-Power Play Is Working

Most APAC regulators do not want to write a full AI law. They want a defensible governance posture without the political cost of a comprehensive statute. Singapore's toolkit gives them that: a vetted open-source framework, international alignment^✦, and an off-the-shelf testing harness. It lowers the political capital needed to act, which is the single biggest blocker in emerging-market regulation.

There is also a sector advantage. The Veritas Toolkit, originally developed by the Monetary Authority of Singapore (MAS) for financial institutions testing against FEAT principles, is now a plugin inside AI Verify. That gives Asian central banks and financial supervisors a ready-made starting point. Bank Negara Malaysia and Bank of Thailand have both indicated internal use.

Singapore does not need to pass an AI law to shape how AI gets deployed in Asia. We are setting the floor and the tools, and the region is picking them up.

Agentic AI changes the risk calculus for every enterprise. Our framework is deliberately modular so it can be adopted in pieces, not swallowed whole.

The Global AI Assurance Pilot

The second piece of the Singapore strategy is the Global AI Assurance Pilot, run jointly with the AI Verify Foundation since February 2025. It connects real-world generative AI deployers with independent testers, codifies test norms, and publishes case studies. That is a subtle but important move: it builds the auditing supply chain that most Asian jurisdictions lack.

Pilot participants have included banks, health providers, and government services from 12 countries across Asia and beyond. The output of the pilot, published in late 2025, has already fed into ISO/IEC 42001:2023 conformance guidance and into the AI Verify 2.0 release.

The Limits Of Soft Power

There is a real ceiling here. A toolkit is not a law. If a Chinese LLM^✦ ships a harmful output in Jakarta, OJK cannot fine it using AI Verify alone. Enforcement still depends on whichever national law applies. So the Singapore approach works best as pre-enforcement governance scaffolding, not as a standalone regulatory regime. That is one reason why Korea, Japan, and India have all chosen to legislate directly rather than adopt a Verify-style toolkit as a primary compliance layer.

It also matters who controls the framework. AI Verify is governed by a Singapore statutory body and a Singapore-led foundation. That concentration is beginning to attract quiet pushback. ASEAN observers have asked for more shared governance of the toolkit, which IMDA has begun to address through the ASEAN Working Group on AI Governance (WG-AI), of which Singapore is the convenor.

The Singapore strategy does not sit in isolation. Read it alongside our reporting on Korea's AI Basic Act enforcement, Vietnam's phased AI law, and the ASEAN-Japan AI co-creation framework. For the private-sector pressure these toolkits now respond to, our look at Malaysia's AI sandbox graduates shows what deployment looks like on the ground.

Frequently Asked Questions

Is AI Verify mandatory for deploying AI in Singapore?

No. It is a voluntary testing framework. Singapore has not passed a mandatory AI law. In practice, financial regulators, healthcare authorities, and government tenders increasingly expect Verify-style attestation as part of due diligence.

How is AI Verify different from the EU AI Act?

The EU AI Act is binding legislation with risk-based prohibitions and fines. AI Verify is a voluntary governance toolkit that maps to multiple international standards. They are complementary rather than competing: a deployer in Europe could use AI Verify to demonstrate conformity with the EU AI Act's technical documentation requirements.

Which Asian jurisdictions have adopted it so far?

Malaysia, Indonesia, and Thailand have each publicly referenced AI Verify in regulatory or advisory guidance. Central banks in Malaysia and Thailand are piloting the Veritas plugin for financial AI testing. More ASEAN adoption is expected in 2026.

Can non-Singapore firms use the framework freely?

Yes. AI Verify is open-source and available via GitHub under a permissive licence. There is no licensing fee, and the AI Verify Foundation offers free onboarding support for qualifying enterprises.

If Singapore's toolkit becomes the de facto Asian AI compliance standard, does the region still need a binding ASEAN-wide AI treaty? Drop your take in the comments below.