The Balancing Act: Harmonising AI Development and Governance in Southeast Asia

Southeast Asia is experiencing a rapid surge in AI development. In McKinsey & Company’s latest study, nearly half of firms in the region are already scaling or have scaled AI beyond the pilot phase—outpacing the global average of 35%.

This momentum is reinforced by Google’s AI Opportunity report, which estimates that AI could contribute up to USD 270 billion to the region’s economy. As adoption accelerates, the need for robust AI governance and safety frameworks becomes increasingly urgent.

However, Southeast Asia’s linguistic and cultural diversity introduces added complexity. Globally trained AI models often struggle with low-resource regional languages and dialects, leading to risks such as bias, toxic outputs, or safety blind spots. This makes localisation of AI safety approaches essential.

At the same time, governance should not be seen as a constraint on innovation. Instead, it is a foundation for trust—ensuring AI systems are developed and deployed responsibly while delivering broad societal value.

From Regional Frameworks to National Interpretation

As AI systems become more powerful and widely adopted, the risks they pose increasingly transcend national borders. According to Google’s AI Opportunity report, 89% of respondents agreed that AI needs to be developed responsibly. AI’s transnational nature makes isolated, fragmented responses insufficient, underscoring the need for a coordinated regional approach to AI governance and safety. At the same time, addressing shared challenges requires navigating uneven levels of institutional readiness across countries.

This is where the Association of Southeast Asian Nations (ASEAN) plays a pivotal role. As a regional platform, ASEAN brings together its eleven member states to facilitate collaboration and promote more cohesive governance.

In 2024, the ASEAN Guide on AI Ethics and Governance was released, followed by the establishment of the ASEAN Working Group on AI Governance (WG-AI), a key recommendation from the guide. In 2025, this was further expanded through the development of the Expanded ASEAN Guide on AI Governance and Ethics, with a specific focus on generative AI.

In parallel with these regional efforts, individual countries have taken steps to strengthen national AI policies and safeguards. Six ASEAN member states have established national AI strategies, while Cambodia and Laos are currently drafting their plans following the completion of their respective UNESCO Readiness Assessment Methodology reports. Brunei, Myanmar, and Timor-Leste remain without specific national AI strategies, although Brunei has published a Guide on AI Governance and Ethics in 2025.

Given the relative nascency of AI development in the region, countries in Southeast Asia largely favour a soft law approach. Countries such as Singapore, Indonesia, Malaysia, and Thailand have issued voluntary guidelines to address AI governance and safety as their ecosystems mature.

• Malaysia: The National Guidelines on AI Governance and Ethics (AIGE) is the current foundational document on AI governance. The guidelines outline rights and responsibilities of different actors within the AI development process based on seven core principles aligned with international guidelines.
• Thailand: Established an AI Governance Clinic (AIGC) via the Electronic Transactions Development Agency (ETDA). The AIGC serves as a platform for knowledge sharing, capacity building, and practical implementation of AI governance policy documents (AI Ethics Guidelines, Guidelines for Generative AI Governance).
• Singapore: The InfoComm Media Development Authority (IMDA) developed a Model AI Governance Framework that has been revised three times since its initial release to account for developments such as generative AI and agentic AI.
• Indonesia: AI governance has been approached through circular letters issued by the Ministry of Communication and Digital Affairs (KOMDIGI), establishing national AI principles. At the sectoral level, the Financial Services Authority (OJK) has introduced guidelines on Artificial Intelligence Governance for Indonesian Banks.

However, the entry into force of Vietnam’s AI Law in March 2026 marks a historic milestone for ASEAN, introducing the region’s first binding legal framework on AI. Vietnam’s approach balances elements of the regulatory-first model seen in the European Union with innovation-led approaches such as those in Korea and Japan.

With Malaysia, Thailand, Indonesia and the Philippines currently proposing AI laws spanning both hard and soft law approaches, Vietnam’s progress may serve as a bellwether for the region’s legislative trajectory. Notably, Singapore remains absent from the AI law conversation, with no indication that a binding framework is forthcoming.

Malaysia – Nationally Rooted with Global Ambitions 

Malaysia’s AI development agenda is anchored in its broader national digital transformation efforts, as set out in the Malaysia Digital Economy Blueprint and the National 4IR Policy. This ambition has been further reinforced by the 13th Malaysia Plan (2026–2030), which positions AI as a key pillar of national development and sets the aspiration of making Malaysia one of the world’s leading AI nations by 2030.

Addressing the governance gap: building national coordination and safeguards

As AI adoption accelerates, Malaysia’s policy focus has shifted towards strengthening governance foundations to ensure that innovation is matched with accountability, trust, and safety.

To support this, the Ministry of Digital established the National AI Office (NAIO) in 2024 as the central coordinating body for AI development. NAIO’s mandate includes driving AI innovation, advancing governance frameworks, and fostering cross-sector collaboration across government, industry, and research ecosystems.

A key initiative under NAIO is the development of an AI Governance Bill, which seeks to translate governance principles into a more structured regulatory framework. The Bill is anchored on six core areas:

1. Embedding foundational AI governance principles into law
2. Assigning clear responsibility across the AI lifecycle
3. Establishing dedicated governance institutions
4. Introducing a risk-based classification framework
5. Creating a structured incident reporting system
6. Launching an AI sandbox to enable responsible experimentation

The Bill, expected to be tabled in Parliament this year, aims to strike a balance between enabling innovation and strengthening accountability, providing a more agile governance structure for Malaysia’s evolving AI ecosystem.

Extending influence regionally: the ASEAN AI Safety Network

Beyond national policy, Malaysia is also shaping the regional AI governance agenda through its leadership of the proposed ASEAN AI Safety Network (ASEAN AI Safe). The initiative has been designated as a Priority Economic Deliverable (PED) under Malaysia’s ASEAN Chairmanship in 2025, underscoring its dual ambition to advance both national and regional governance capacity.

This regional focus is particularly important given that AI safety remains a relatively nascent area across Southeast Asia. The region’s multilingual and multicultural context also means that AI systems must be evaluated and governed with local realities in mind.

At present, technical AI safety capabilities—such as red-teaming, benchmarking and evaluation, alignment research, and threat modelling—remain limited compared to more established ecosystems such as the United States, United Kingdom, Korea, and Japan. The ASEAN AI Safe therefore presents an opportunity to address these gaps through a more coordinated regional approach.

Following the adoption of the Declaration on the Establishment of an ASEAN AI Safety Network at the 47th ASEAN Leaders’ Summit in October 2025, Malaysia is currently establishing the operational secretariat for ASEAN AI Safe, which will be based in the country.

Once operational, the ASEAN AI Safe will serve as a regional coordination platform to strengthen institutional capacity for AI safety across ASEAN. It will support the ASEAN Digital Senior Officials’ Meeting (ADGSOM) and the ASEAN Working Group on AI Governance (WG-AI) by enabling knowledge exchange, building technical capacity, and deepening collaboration among member states.

Conclusion

While global conversations have already shifted towards addressing frontier and existential risks of agentic AI, it is vital that Southeast Asian countries, home to some of the world’s most dynamic digital economies, are represented in these discussions to ensure that the region’s unique perspectives and risk profiles reflective of its multicultural and multilingual societies are adequately addressed. Malaysia hopes to contribute to the region’s development via the establishment of the ASEAN AI Safe, supporting fellow ASEAN member states to build solid governance and safety foundations that reflect their respective policy and operating realities. As the region continues its AI journeys, good governance and robust safety capabilities provide the essential guardrails for the sustainable and responsible development and deployment of AI.

The views and recommendations expressed in this article published on are solely of the authors and do not necessarily reflect the views and position of the Tech for Good Institute.

Read the second part of this series, where we take a deeper look at the important work of the ASEAN AI Safety Network.