Governing AI the Asean way

THE acceleration of artificial intelligence (AI) in 2025 was remarkable.

AI within Asean moved from aspirational principles into a much more coordinated mission where expanded AI ethics guidelines, responsible AI roadmaps and closer policy alignment reflected a growing consensus.

We learnt that innovation thrives when trust, safety and human interests are clearly in the frame.

Early in the year, Asean digital ministers endorsed an expanded Guide on AI Governance and Ethics alongside a Responsible AI Roadmap.

In fact, Malaysia is among the early movers in Asean to publish national AI governance guidelines, organised across three tiers – users, regulators and developers.

The year’s standout moment came under Malaysia’s Asean chairmanship, with the launch of the Asean AI Safety Network – a signal that the region was serious about moving from good intentions to shared guardrails.

We also saw rapid ecosystem growth within the region, notably the  Asean-China Week 2025 in Fuzhou, Fujian which focused on cooperation between China and Asean on technological development and governance.

Why does all this matter really?

Because innovation at scale depends not just on technology readiness but on regulatory confidence and trust.

Simply put, by prioritising interoperable, innovation-friendly standards over a one-size-fits-all law, Asean enables cross-border AI deployment while carefully balancing speed, responsibility and national sovereignty.

 

Navigating the gap

Malaysia has positioned itself well.

I spoke earlier about the country being one of the first movers and that’s not all it has done.

Digital Minister Gobind Singh Deo announced the Asean AI Safety Network will be headquartered in Kuala Lumpur, reinforcing the country’s commitment to trusted regional digital governance.

This also means Malaysia has the opportunity to take hold of the reins and steer the ship towards stronger AI governance within the region.

It is interesting to observe where other countries are on their respective journey and the stages they have reached.

Now, we know that countries like the US have relied on executive orders and agency guidelines for governance matters.

By contrast, the EU AI Act (the first comprehensive regulation on AI by a major regulator) which went into effect in August 2024 has enforcement mechanisms, scientific panels and national authorities.

While the US promotes innovation-first self-governance, the EU has codified ethics into law.

Divergent AI governance approaches may be well-intentioned, but without coordination they risk fragmenting and slowing progress.

I posit that there is a clear vacuum where AI governance can sit and Asean can play a critical role in filling it.

One might argue that the EU’s risk-based model offers structure, while the US’s emphasis on flexibility provides pathways better suited to Asean’s diversity.

Regardless, there is much to learn from both.

One thing I’ve noticed is, the EU has a tiered model where it acts on risks, so even if a certain system becomes particularly sensitive or risky, it must comply with certain requirements.

Transparency is another big thing within the EU and one that we can follow.

The EU is moving toward obligations to disclose training data sources, respect copyright optouts and label AI-generated content to tackle misinformation and protect creators.

What about the US, what lessons can we take from them?

Well, for one – they preserve plenty of space for experimentation, more commonly known as sandboxes.

Using a controlled and carefully monitored environment, AI can safely be tested out without putting any person or a market at risk.

This is actually a really good idea and has been used in financial tech and in universities across the US.

Another smart aspect of the US way of governing AI is their sheer level of competitiveness – US policy frames AI governance alongside innovation, supply chains and security.

This includes controlling exports to protect AI hardware and software stacks from adversaries and something Asean can emulate so we, as a region remain independent.

Unfortunately, at its current state, AI governance in the region is fragmented at its best.

There are a number of reasons for this, but mostly because of the lack of enforcement. While the existing framework functions as an advisory document, deterrence power is almost non-existent.

Then we also see a significant divide in national readiness, as countries like Singapore are far ahead of say, Myanmar.

 

Lesson to take onboard?

A regional governance framework must reflect differences in institutional capacity and infrastructure to be both effective and inclusive.

While the region wants to remain sovereign, it is indeed wise to still engage with international blocs.

The EU actively participated in leading international platforms like G7, G20, OECD and the Global Partnership on AI among others.

What’s important to remember is, AI is global by nature, but local by design.

At TM R&D, we began experimenting with large language models (LLM) in late 2022, laying the groundwork for a generative AI–powered interactive chatbot. Unlike traditional rule-based chatbots, it uses fine-tuned LLMs designed for specific domains. Guardrails are built in to keep responses accurate and within context, reducing the risk of misinformation while enabling more natural and meaningful interactions.

Building on this experience, an Asean-built LLM could support regional needs in education, healthcare and sustainability, while ensuring AI remains relevant and inclusive across Asean.

In addition, TM R&D is championing the development of a quantum network lab to create a network safe infrastructure for Asean to secure LLM infrastructure and to protect data in transit with post-quantum cryptography (PQC) algorithms.

Above all, TM R&D contributes to AI governance by embedding responsible, secure and compliant AI practices into research, systems and real-world deployments, in line with Malaysia’s Personal Data Protection Act (PDPA).

 

From principles to practice

As we look ahead to 2026, the next phase needs to be defined less by ambition and more by action.

There are already shared principles pertaining to responsible and inclusive AI, the challenge will be to turn these into interoperable mechanisms that can be adopted by every member country in the region.

Concerns are real, generative AI misuse – from deepfakes to automated disinformation are testing existing regulatory tools.

Concurrently, cross-border data flows are expanding rapidly, raising questions around sovereignty, privacy and trust.

Against this backdrop, Asean’s influence in AI governance will not hinge on replicating the EU’s rules-heavy approach or the US’s market-led model.

Instead, the region’s strength lies in its ability to craft a distinct, trust-based framework that reflects regional realities.

This means investing in regulatory capacity, strengthening cross-border cooperation, and creating practical guardrails that can evolve alongside technology.

If Asean can strike this balance – firm where it matters, flexible where it counts, it will not only manage AI risks but also position the region to harness AI as a driver of sustainable growth and shared prosperity.

---

This article was first published in the StarBiz 7, Issue Feb 7, 2025: People first in healthcare reset

To access the PDF version, Click here

To read the full Star epaper: Click here