Across Southeast Asia (SEA), governments are treating AI as critical national infrastructure, focusing heavily on strategic investment and sovereign AI. For instance, Thailand has launched a home-grown sovereign AI infrastructure designed specifically for the Thai language and culture. ThaiLLM has been trained on over 100 billion tokens of high-quality Thai data, including government documents, educational research, and legal texts. Simultaneously, governance frameworks are maturing alongside infrastructure. As an example, Malaysia’s AI Governance Bill which is being drafted, will cover the full AI lifecycle, forming a “preventive layer” to complement existing statutes. In addition, SEA governments’ conscious decision to invest in human and research capacity is an acknowledgement that the harnessing of AI is not only about the technology, but whether people have the right skills to take advantage of its benefits.
Against this backdrop, the Tech for Good Institute (TFGI) facilitated a workshop with senior-level business representatives from the region on navigating the promises and risks that AI would bring. Participants were challenged with questions surrounding the key compliance and cross-border risks they are facing in the time of AI governance. In addition, participants also shared what governments in the region can do to promote AI use, while safeguarding the digital society from unintended consequences.
Facilitators
- Keith Detros, Programme Manager, Tech for Good Institute
- Fairoz Ahmad, Programme Fellow, Tech for Good Institute
Key Takeaways
- The 3S: Safety, Sovereignty, and Skilling
Through TFGI’s roundtables across the SEA-6, three key pillars emerged: safety, sovereignty, and skilling. Safety serves as the foundation for AI adoption, moving away from rigid mandates towards flexible, outcome-based approaches. For instance, the Philippines formulated a National AI Governance Framework to ensure sound governance, AI-ready data systems, enhanced capabilities, and expanded infrastructure. Meanwhile, sovereignty means different things to different countries. Singapore for example focuses on operational oversight, leveraging regional interoperability to complement national control. In Malaysia, the government has earmarked approximately USD 460 million for a National Sovereign AI Cloud, scheduled for launch in 2026, with the aim of positioning the country as a regional cloud and digital hub by 2030. Lastly, skilling allows for the effective scaling of AI within the workforce. Singapore has taken the path of enterprise-led adoption by supporting small businesses in adopting AI. Its National AI Impact Programme (NAIIP) will strengthen Singapore’s AI capabilities across enterprises and workers, supporting 10,000 enterprises over the next three years to advance their AI adoption journey and helping 100,000 workers become AI Bilingual.
- Data sovereignty and security as the top compliance issue
Data sovereignty and security emerge as a key concern for participants, which reflects the tension between operational efficiency and robust data governance. The debate focused on understanding the risks of exposing data to third parties, which carries significant long-term investment implications. This tension is most salient when businesses collect customer data in one country but the data centre is based in a neighbouring country. In addition, when customer data is routed through a jurisdiction with different data protection laws, questions are bound to arise regarding application of data governance regimes.
Consequently, without comprehensive AI laws across SEA-6 markets, businesses remain governed by existing data privacy and security statutes that are inadequate to address AI-specific risks. This results in a fragmented regional compliance landscape. For example, Vietnam has advanced the furthest in this regard, becoming the first in the region to pass a legally binding AI Law featuring a four-tier risk classification system. Nonetheless, participants stressed the urgent need for cross-jurisdictional alignment as AI infrastructure increasingly operates across borders.
- Deskilling risks in AI-driven workforce transformation
Micro, Small, and Medium Enterprises (MSMEs) make up 97% of ASEAN businesses and 85% of regional employment. This statistic reveals a worrying trend as 43% of employers expect core skills to be disrupted within five years, resulting in skills mismatch and job displacements. The region is responding to this shift in the job market by implementing domestic policies. To highlight one, Malaysia is empowering local businesses to adopt emerging technologies such as blockchain, AI, and quantum computing through the MYR 53 million Malaysia Digital Acceleration Grant. s. An example shared by participants cited the loss of ability for specialist junior doctors to diagnose conditions with the aggressive roll out of AI diagnostic imaging. Thus, when AI adoption across different industries is pursued without providing workers opportunities to learn, the workforce could face long-term challenges in training early-career professionals adequately.
