Southeast Asia’s approach to AI governance is visibly hardening in 2026. What used to be a patchwork of voluntary frameworks and sector-specific advisories is consolidating into more structured, in some cases legally binding, regimes, and the Philippines has not yet made its own move.
Vietnam is the clearest example of the shift. It became the first country in the region to pass a concrete, binding AI law, with the legislation taking effect in a phased rollout starting March 2026 and continuing over four years. That timeline gives businesses room to adjust, but it establishes Vietnam as the first Southeast Asian jurisdiction to move beyond guidance and into enforceable statute.
Singapore has taken a different route. It enters this period without a standalone AI law, continuing to rely on voluntary tools like its AI Verify testing framework and sector-specific guidance. But in January 2026, Singapore’s Infocomm Media Development Authority launched what it describes as the world’s first Model AI Governance Framework specifically for agentic AI, the class of systems that can take autonomous, multi-step actions rather than simply respond to prompts. Given how quickly agentic tools are being embedded into everything from coding assistants to financial workflows, Singapore’s early move here is likely to become a reference point for how other regulators eventually define agentic-AI accountability.
Elsewhere in the region, Malaysia and Brunei have largely stuck with voluntary risk frameworks rather than binding rules, and there is an active regional debate, tracked by outlets like Modern Diplomacy, over whether ASEAN as a bloc should push toward harmonized binding rules or continue letting individual member states set their own pace.
The Philippines currently has no AI-specific statute. Existing coverage comes indirectly, through the Data Privacy Act for data-handling obligations and through the general innovation-support framework of the Innovative Startup Act, Republic Act No. 11337, neither of which was written with frontier AI models or agentic systems in mind. That gap is becoming harder to ignore for two reasons converging at once: a reported $10 billion US-backed AI infrastructure push at Clark Freeport Zone that will draw international scrutiny of the country’s regulatory maturity, and a domestic fintech sector, GCash and Maya chief among them, already experimenting with AI-driven fraud detection and customer service that regulators will eventually need clearer rules for. Vietnam’s binding law and Singapore’s agentic-AI framework both give the Philippines a template to borrow from; the question is whether DICT, DOST, and DTI move before foreign capital and regional peers effectively set the bar first.
The regional debate over binding versus voluntary rules is not simply a legal technicality. Analysts tracking the shift describe Southeast Asian governments as borrowing explicitly from the European Union’s risk-based regulatory playbook, tiering obligations by how much potential harm a given AI application could cause rather than regulating all AI use cases uniformly. Hong Kong and Singapore, despite taking a lighter regulatory touch than Vietnam, are reportedly running a similar mid-year policy checkpoint process to decide whether their voluntary frameworks need to harden into something closer to binding law within the next year or two, suggesting the region as a whole is converging toward stricter rules even where individual governments are moving at different speeds.
Share this article