India, Japan, South Korea & Singapore: Alternative Asian Models

Much of the global AI governance debate is framed around three major models.

Europe is associated with rights-based, risk-oriented regulation.

The United States is associated with innovation, national security, standards and distributed governance.

China is associated with state strategy, regulation, cybersecurity, industrial policy and technical standards.

But this three-model framework is incomplete.

Some of the most interesting developments in AI governance are occurring elsewhere in Asia.

India, Japan, South Korea and Singapore are each developing distinctive approaches to artificial intelligence.

None of these countries simply reproduces the European, American or Chinese model.

Instead, they are experimenting with different combinations of innovation policy, national AI infrastructure, safety, standards, regulation, skills development, public-sector adoption and international cooperation.

This makes them particularly important for understanding the future of AI governance.

It also makes them highly relevant for developing countries.

For countries that are neither frontier AI superpowers nor large regulatory blocs, the Asian experience offers a different question:

How can a country build AI capability while simultaneously developing credible governance?

The answer appears to vary considerably.

India: AI Governance Through Capability and Inclusion

India’s approach to AI governance is closely connected to national development and technological capability.

Rather than beginning with a comprehensive AI regulatory statute equivalent to the EU AI Act, India has built a broader policy ecosystem around the IndiaAI Mission.

The Government of India approved the IndiaAI Mission in March 2024 as a national programme designed to build a comprehensive AI ecosystem.

Its seven pillars cover areas including compute capacity, datasets, indigenous foundation models, applications, future skills, startup financing and safe and trusted AI.

This is significant.

India’s AI governance approach is not being developed separately from AI industrial policy.

Compute is part of AI policy.

Data is part of AI policy.

Skills are part of AI policy.

Startup development is part of AI policy.

Indigenous AI capability is part of AI policy.

And safe and trusted AI is part of the same national programme.

This creates a governance model centred strongly on capacity.

The IndiaAI Mission

The structure of the IndiaAI Mission illustrates this clearly.

The programme seeks to expand access to computing resources, improve datasets, support domestic AI capabilities, develop applications, build skills and provide financing while also supporting responsible AI.

IndiaAI’s compute programme is designed to provide access to AI computing resources to researchers, academia, startups, government entities and other approved users. The programme’s policy documents explicitly frame compute access as part of India’s effort to strengthen AI competitiveness and democratise access to infrastructure.

This is an important governance insight.

A country cannot effectively govern AI if it lacks the capacity to understand or participate in AI development.

Governance therefore begins partly with capability.

This is particularly relevant for developing countries.

If advanced AI systems are developed entirely elsewhere, countries may become primarily consumers of externally developed technology.

They may have limited ability to evaluate models independently.

They may depend on foreign infrastructure.

They may have limited access to technical expertise.

India’s approach attempts to reduce some of these dependencies by investing directly in national AI infrastructure and talent.

AI Governance as Development Policy

India’s approach also places strong emphasis on inclusion.

The official IndiaAI Mission describes its objectives in terms of democratizing access to AI, developing domestic expertise and promoting socially impactful applications.

This creates an important connection between AI governance and development policy.

For India, AI governance is not only about preventing harmful uses.

It is also about ensuring that AI contributes to economic development and public value.

This includes healthcare, education, agriculture, public administration and other sectors.

Such an approach is particularly relevant for the Global South because governance questions often emerge from development constraints.

A country may be more concerned about access to AI-enabled healthcare than about regulating a frontier model’s abstract capabilities.

It may need AI for agricultural productivity, education or public services while simultaneously dealing with limited regulatory and technical capacity.

This does not make safety less important.

It changes the governance context.

The 2025 India AI Governance Guidelines

India’s governance framework has continued to develop.

In November 2025, India’s Ministry of Electronics and Information Technology introduced India AI Governance Guidelines under the IndiaAI Mission, framing them around safe, inclusive and responsible AI. Official IndiaAI materials describe the guidelines as part of India’s emerging governance approach.

This suggests that India is moving toward a model in which governance guidance is developing alongside national AI capability.

The sequence is important.

India has not simply waited for a comprehensive regulatory statute.

It has built infrastructure, skills, datasets and innovation capacity while developing governance mechanisms.

This is a different path from the European approach.

The question is not only:

How should AI be restricted?

It is also:

How can the country build the capacity necessary to participate in AI while establishing safeguards?

Japan: Innovation With Institutional Coordination

Japan provides another distinctive model.

Its approach has historically relied heavily on guidelines, voluntary governance and coordination between government and industry.

Japan’s AI policy developed partly through the 2019 Social Principles of Human-Centric AI and later AI Guidelines for Business.

The approach emphasised human-centred AI, safety, transparency, accountability and responsible development while avoiding a highly restrictive regulatory framework.

Japan’s position changed institutionally in 2025 when it enacted the Act on Promotion of Research and Development and Utilization of Artificial Intelligence-Related Technologies.

The law was promulgated in June 2025 and fully entered into force in September 2025.

But the Japanese model remains different from the EU AI Act.

The law is primarily designed to promote AI research, development and utilisation while addressing risks.

It establishes an AI Strategic Headquarters chaired by the Prime Minister and requires an AI Basic Plan.

It also provides for government guidelines, information gathering and investigation of problematic AI-related incidents.

The Japanese approach therefore combines legal institutionalisation with relatively flexible governance mechanisms.

The State Coordinates Rather Than Classifies Everything

This is one of the most important characteristics of Japan’s model.

The Japanese AI Act does not simply reproduce the EU’s extensive risk-classification architecture.

Instead, it creates national coordination structures and gives government responsibility for developing comprehensive AI policy.

Official Japanese government material describes the framework as combining existing laws with guidelines designed to respond to emerging AI risks while avoiding excessive regulation that could slow innovation.

This creates a governance model in which flexibility is important.

Rather than attempting to predict every future AI use through legislation, the government can issue and update guidance as technology changes.

Japan’s approach is therefore particularly relevant to the question of regulatory adaptability.

AI Safety and the Hiroshima Legacy

Japan also occupies an important position in international AI governance because of the Hiroshima AI Process discussed earlier in this series.

Japan’s domestic AI governance has therefore developed alongside its international role in AI safety diplomacy.

The Japanese AI Act itself draws on the Hiroshima AI Process as part of its broader policy context.

This creates an interesting relationship between domestic and international governance.

International diplomacy can influence domestic policy.

Domestic institutions can support international commitments.

And international principles can become reference points for national guidelines.

This is an important lesson for countries that participate in international AI governance forums.

International engagement is most meaningful when it is connected to domestic institutional capacity.

Japan and AI Infrastructure

Japan’s AI strategy also increasingly connects governance with infrastructure.

In February 2025, the Japanese government adopted legislation aimed at supporting advanced semiconductors and AI-related technologies, recognising growing demand for computation associated with generative AI and the need to develop an ecosystem connecting hardware, data centres and AI software.

This reflects a broader global trend.

AI governance is becoming inseparable from compute.

A country cannot build an effective AI strategy without considering chips, data centres, energy, research infrastructure and skilled people.

Japan’s experience demonstrates how AI policy can therefore become part of industrial and infrastructure policy.

South Korea: AI Regulation Meets Industrial Strategy

South Korea represents another important case.

The country has developed an approach that combines AI promotion with a dedicated legal framework for safety and trust.

The Basic Act on the Development of Artificial Intelligence and the Establishment of a Trustworthy Foundation was passed by the National Assembly in December 2024 and took effect on 22 January 2026.

South Korea described the legislation as a foundation for both national AI competitiveness and trustworthy AI utilisation.

This is an important combination.

The law does not treat industrial development and governance as separate agendas.

It connects them.

The AI Basic Act

The Korean framework establishes a national AI governance structure and includes provisions addressing high-risk AI and generative AI.

It creates obligations relating to transparency, safety and operator responsibilities, while also providing a legal basis for voluntary safety and trustworthiness verification and AI impact assessments.

This makes South Korea an interesting hybrid.

It has a dedicated AI law, unlike the more fragmented US approach.

But the law is also strongly connected to industrial development and national AI competitiveness.

It therefore sits somewhere between comprehensive regulation and development-oriented AI policy.

Implementation in 2026

The Korean framework became particularly important in 2026 because the AI Basic Act entered into force on 22 January.

The government developed subordinate regulations and guidelines during the preparation period and established a grace period of at least one year for certain enforcement measures.

The Ministry of Science and ICT also issued guidelines concerning AI transparency, including implementation measures and labelling requirements for AI-generated content.

This demonstrates a pattern increasingly visible across AI governance.

Passing the law is only the beginning.

The real governance system emerges through enforcement guidance, standards, institutional capacity and implementation procedures.

South Korea’s experience is therefore useful for studying the transition from legislation to operational governance.

Safety and National Capability

South Korea has also been developing AI safety capabilities alongside regulation.

The government announced plans for a national AI Safety Research Center and emphasised AI safety, security and international cooperation as components of its broader AI strategy.

This is significant because it reflects a wider shift from policy principles toward technical capacity.

A country needs institutions capable of evaluating AI systems.

It needs researchers capable of studying safety.

It needs standards.

It needs testing.

It needs public-sector expertise.

This is the same underlying problem identified by the Bletchley and Seoul processes.

The governance question increasingly becomes:

Who has the technical capacity to verify AI safety?

Singapore: Governance Through Implementation and Assurance

Singapore offers perhaps the clearest example of a different Asian approach.

Rather than beginning with comprehensive binding legislation, Singapore has developed practical governance frameworks, testing tools, standards-oriented initiatives and close cooperation with industry.

The country released its first Model AI Governance Framework in 2019 and a second edition in 2020.

The framework was designed to translate broad AI ethics principles into practical organisational measures.

This made Singapore an important example of implementation-oriented governance.

The key idea is practical governance.

Organisations need to know what responsible AI means in practice.

They need internal governance structures.

They need risk-management processes.

They need appropriate human involvement.

They need transparency.

They need ways to test AI systems.

Singapore therefore moved beyond principles relatively early.

AI Verify

One of Singapore’s most important governance innovations is AI Verify.

Launched in 2022, AI Verify is a governance testing framework and toolkit designed to help organisations demonstrate responsible AI through technical tests and process checks.

This is strategically important.

Singapore’s model asks a practical question:

How can organisations demonstrate that an AI system meets governance expectations?

That is fundamentally an assurance question.

It moves governance from:

“We believe this AI is responsible.”

toward:

“Here is evidence that we tested relevant properties of the system.”

This distinction will become increasingly important as AI regulation becomes more technical.

Generative AI Governance

Singapore extended its framework to generative AI in 2024.

The proposed Model AI Governance Framework for Generative AI addressed emerging risks while maintaining a focus on innovation and international interoperability. It drew on technical testing work and sought to provide practical guidance to developers and policymakers.

This is another distinctive characteristic.

Singapore is attempting to build governance tools that can work across jurisdictions.

Its AI Verify framework has been designed with reference to international approaches, including frameworks associated with the EU, United States and OECD.

This gives Singapore a potentially important role as an interoperability hub.

Instead of attempting to impose one national model globally, it can help connect different governance systems through standards and testing.

The Agentic AI Challenge

Singapore’s approach has become particularly relevant in 2026 because it has moved into agentic AI governance.

In January 2026, Singapore launched a Model AI Governance Framework for Agentic AI.

The framework focuses on risks created by AI systems that can plan and take actions on behalf of users.

It recommends limiting agent autonomy and access to tools and data, defining meaningful human approval checkpoints, implementing technical controls throughout the agent lifecycle and educating end users.

In May 2026, the framework was updated with real-world case studies and additional practices addressing multi-agent systems, third-party agents and automation bias. The update incorporated feedback from more than 60 organisations.

This is significant for the future of AI governance.

Singapore is moving from governing AI systems that primarily generate outputs toward governing systems that can take actions.

That changes the governance problem.

When an AI agent can send an email, modify a database, execute code, access an API or initiate a transaction, the consequences of failure become more direct.

Governance therefore needs to address autonomy, permissions, reversibility and human intervention.

Singapore’s approach is an early example of this emerging governance problem.

Four Asian Models, Four Different Priorities

Looking across India, Japan, South Korea and Singapore reveals important differences.

India places strong emphasis on national AI capability, infrastructure, inclusion and development.

Japan combines innovation promotion with government coordination, guidelines and increasingly formal legal institutions.

South Korea combines a dedicated AI law with industrial development, safety, transparency and national competitiveness.

Singapore focuses heavily on practical governance, testing, assurance, interoperability and implementation.

These differences are important.

There is no single “Asian AI model.”

Asia contains multiple governance approaches.

Some are more infrastructure-focused.

Some are more regulatory.

Some emphasise voluntary governance.

Some emphasise national industrial strategy.

Some focus strongly on technical assurance.

And many combine several of these approaches.

The common thread is that AI governance is being developed alongside national AI capability.

This is perhaps the most important lesson.

Governance and Development Are Connected

For developing countries, this is particularly relevant.

AI governance cannot be separated completely from questions of development.

A country needs technical talent to evaluate AI.

It needs universities to conduct research.

It needs infrastructure to experiment.

It needs public-sector capacity to use AI responsibly.

It needs regulatory institutions to address harmful uses.

It needs standards and testing mechanisms.

And it needs international cooperation to avoid becoming isolated from global governance developments.

India’s approach demonstrates the importance of national infrastructure.

Japan demonstrates the value of institutional coordination.

South Korea demonstrates the possibility of combining legislation with industrial policy.

Singapore demonstrates how technical assurance and interoperability can become governance infrastructure.

These are valuable lessons for countries that do not have the resources of the largest AI powers.

The Global South Question

This is where Asian governance models become especially important for the Global South.

The traditional AI governance debate has often been dominated by Europe, the United States and China.

But many countries face different circumstances.

They may have smaller AI research ecosystems.

They may depend on foreign cloud infrastructure.

They may have limited access to advanced chips.

They may have smaller regulatory institutions.

They may face shortages of AI safety expertise.

They may need AI primarily for development applications.

They may have linguistic and cultural requirements that are underrepresented in global AI systems.

For these countries, the question is not simply:

Which major-power regulatory model should we copy?

The more useful question is:

Which governance functions are essential, and how can they be developed within our own institutional capacity?

This is where the Asian experience becomes particularly valuable.

Governance as Capacity Building

One of the strongest common themes across these countries is capacity building.

AI governance increasingly requires:

technical expertise,

standards,

evaluation,

research,

infrastructure,

skills,

public-sector capability,

institutional coordination,

and mechanisms for international cooperation.

This suggests that AI governance should not be understood only as a regulatory burden.

It can also be understood as a form of national capability.

A country that can independently evaluate AI systems has more governance autonomy than a country that must rely entirely on external providers.

A country with AI standards expertise has greater influence in international discussions.

A country with strong university research capacity can adapt governance frameworks to local conditions.

A country with public-sector AI expertise can deploy systems more responsibly.

This is particularly important for Bangladesh.

What These Asian Models Mean for Bangladesh

Bangladesh sits in a very different institutional and economic position from India, Japan, South Korea and Singapore.

It therefore should not attempt to reproduce any of these systems directly.

But Bangladesh can learn from their institutional functions.

India raises the question of national AI infrastructure, datasets, skills and inclusive AI development.

Japan raises the question of how to combine innovation with flexible governance and national coordination.

South Korea demonstrates how AI legislation can be connected with industrial development and national competitiveness.

Singapore demonstrates the value of AI testing, assurance, practical governance frameworks and international interoperability.

These lessons could be studied systematically.

For Bangladesh, a useful AI governance architecture might eventually need several layers.

There would need to be a national policy framework.

There would need to be sector-specific governance.

There would need to be technical standards.

There would need to be AI impact assessment.

There would need to be institutional responsibility.

There would need to be testing and evaluation capacity.

There would need to be university research.

And there would need to be mechanisms for participating in international AI governance.

But the sequence matters.

A country with limited technical capacity cannot simply import complex regulatory structures and expect them to function automatically.

Institutional development must accompany regulation.

This is one of the most important lessons from the Asian cases.

The Emerging Asian Governance Pattern

The experiences of India, Japan, South Korea and Singapore suggest that a new category of AI governance is emerging.

It could be called capability-oriented governance.

The objective is not simply to regulate AI.

It is to build the national capacity required to develop, deploy, evaluate and govern AI.

This creates a broader governance equation:

AI capability + institutional capacity + safety + development.

That equation may be particularly important for middle-income and developing countries.

The future of AI governance will not be determined only by the countries that develop the largest models.

It will also be shaped by countries that develop effective institutions for using and governing those models.

From Regulation to Implementation

Another important conclusion is that the Asian examples reinforce a broader trend already visible in the EU, United States and China.

AI governance is moving from principles toward implementation.

India is building infrastructure and governance capacity.

Japan is institutionalising national AI coordination.

South Korea is implementing a dedicated AI law.

Singapore is developing testing and assurance frameworks for generative and agentic AI.

The direction is clear.

The governance conversation is becoming increasingly operational.

The future questions are not only about values.

They are about:

Who evaluates?

Who certifies?

Who monitors?

Who sets standards?

Who trains regulators?

Who provides compute?

Who develops national AI expertise?

Who is responsible when AI systems act autonomously?

These questions bring us to the next phase of this series.

The next challenge is not simply designing AI principles.

It is implementing them.

The Governance Gap

This creates what may become one of the most important concepts in AI policy:

the governance gap.

A country may have AI principles but no evaluation capacity.

It may have a strategy but no implementation institution.

It may have a law but no technical expertise.

It may have AI systems but no auditing capability.

It may participate in international forums but lack domestic research capacity.

The distance between policy ambition and institutional capability is the governance gap.

The Asian experience demonstrates that closing this gap requires more than legislation.

It requires infrastructure, people, standards, institutions and continuous learning.

For countries such as Bangladesh, this may be one of the most important AI policy challenges of the coming decade.

The Next Stage

The evolution of global AI governance has now moved through several stages.

First came the recognition that AI creates societal and ethical risks.

Then came international principles.

Then came generative AI and frontier AI.

Then came international AI safety diplomacy.

Then came comprehensive regulation.

And now the focus is increasingly shifting toward implementation.

How are rules actually enforced?

How are AI systems evaluated?

Who performs audits?

What evidence should organisations provide?

How can regulators verify compliance?

How can countries build the technical institutions required for these tasks?

These questions lead directly into the next part of the series.

The future of AI governance may depend less on who can write the most sophisticated principles and more on who can build the institutional machinery needed to make those principles work.

That is where AI governance enters its next phase:

from principles to implementation.