AI is actively changing everything from how founders code software to how governments deliver public services. But while innovation moves fast, regulations often don’t. Across Southeast Asia, we now face a pivotal moment: How can we design an AI playbook that not only protects the public but also empowers local innovation?
As a technology lawyer who regularly deals with emerging companies, I have witnessed firsthand how policymakers usually struggle to strike a balance between enabling growth and mitigating risks. If we want AI regulation to be both relevant and effective for Southeast Asia, we need to get a few key principles right.
Use “soft law” before “hard law”
Southeast Asia’s startup ecosystems are still evolving. Imposing punitive AI laws too early may drive innovation underground or offshore. Instead, governments can first adopt soft law tools.
The approaches taken by Singapore and Malaysia are notable as they include national AI principles, voluntary codes of conduct, and model governance toolkits serve as examples of how countries can use soft law tools to encourage AI adoption.
These approaches create legal clarity without immediately introducing penalties. They also allow startups to adopt “ethical by design” practices which is especially important for founders looking to expand into more regulated jurisdictions later.
Build institutions that can keep up with AI
Even the best written law may be meaningless without capable institutions to monitor, enforce, and evolve with the emerging technologies. Governments must invest in up-skilling regulators, judges, and public servants in AI literacy.
In Malaysia, the government has formed The National AI Office (NAIO) to drive Malaysia’s AI agenda, and assist with other agencies and relevant stakeholders in establishing a path towards making Malaysia an AI driven economy.
Start with high risk AI use cases
Not all AI is the same. A chatbot that suggests playlists has a different risk profile different from an algorithm that decides whether a person gets a loan approved or a longer prison sentence. Governments may wish to prioritise regulating high risk use cases affecting fundamental rights like privacy, employment, or access to justice as areas that may require guardrails to prevent misuse.
Focusing early regulatory energy on these sensitive domains prevents overreach. It also avoids drowning founders in compliance burdens when they’re building low risk applications. AI regulations should be proportionate, risk based, and scalable
Don’t copy paste Western models
The temptation to adopt existing laws from other continents such as the EU style AI Act laws may be understandable, but may not always be wise. Southeast Asian countries are diverse.
What may likely work in Singapore might not even be suitable for Indonesia or even Vietnam. A blanket regulatory framework risks stifling innovation, especially in smaller digital economies.
In the past few years, the approach by policymakers has been to build principle-based guidelines that are adaptable to their country’s regulatory maturity and digital readiness. In Malaysia, The Artificial Intelligence Governance and Ethics Guidelines (AIGE) was launched in 2024 that sets out seven key principles of AI to voluntary adopt responsible and ethical AI practices.
Additionally, the ASEAN Guide on AI Governance and Ethics also serves as a useful reference for entities in the ASEAN region seeking to adopt AI in commercial applications. The document sets out proposed governance framework, national and regional recommendations, real life use cases, and AI risk impact assessment template. The Expanded ASEAN Guide on AI Governance and Ethics – Generative AI looks at risks of Generative AI and recommends a range of policy recommendations for its responsible adoption.
Align with core national values and constitution
In Malaysia, in addition to the Federal Constitution, we also have the Rukunegara, a set of national principles that underpin the country’s values. In Malaysia, NAIO (of which I am a member under the AI Governance and Ethics Working Group) is examining the local context that may be relevant to operationalising the country’s AI governance models and exploring practical, meaningful ways to implement AI ethically.
Start with high risk AI use cases
Not all AI is the same. A chatbot that suggests playlists has a different risk profile different from an algorithm that decides whether a person gets a loan approved or a longer prison sentence. Governments may wish to prioritise regulating high risk use cases affecting fundamental rights like privacy, employment, or access to justice as areas that may require guardrails to prevent misuse.
Focusing early regulatory energy on these sensitive domains prevents overreach. It also avoids drowning founders in compliance burdens when they’re building low risk applications. AI regulations should be proportionate, risk based, and scalable
Don’t copy paste Western models
The temptation to adopt existing laws from other continents such as the EU style AI Act laws may be understandable, but may not always be wise. Southeast Asian countries are diverse.
What may likely work in Singapore might not even be suitable for Indonesia or even Vietnam. A blanket regulatory framework risks stifling innovation, especially in smaller digital economies.
In the past few years, the approach by policymakers has been to build principle-based guidelines that are adaptable to their country’s regulatory maturity and digital readiness. In Malaysia, The Artificial Intelligence Governance and Ethics Guidelines (AIGE) was launched in 2024 that sets out seven key principles of AI to voluntary adopt responsible and ethical AI practices.
Additionally, the ASEAN Guide on AI Governance and Ethics also serves as a useful reference for entities in the ASEAN region seeking to adopt AI in commercial applications. The document sets out proposed governance framework, national and regional recommendations, real life use cases, and AI risk impact assessment template. The Expanded ASEAN Guide on AI Governance and Ethics – Generative AI looks at risks of Generative AI and recommends a range of policy recommendations for its responsible adoption.
Align with core national values and constitution
In Malaysia, in addition to the Federal Constitution, we also have the Rukunegara, a set of national principles that underpin the country’s values. In Malaysia, NAIO (of which I am a member under the AI Governance and Ethics Working Group) is examining the local context that may be relevant to operationalising the country’s AI governance models and exploring practical, meaningful ways to implement AI ethically.
A version of this article was first published on e27.