Han-Wei Liu (Singapore Management University, Yong Pung How School of Law) has posted Global Norms, Diverse Paths: AI Governance in ASEAN (Penn Asian Law Review, forthcoming) on SSRN. Here is the abstract:
Comparative scholarship on AI governance has concentrated overwhelmingly on the European Union, the United States, and China. ASEAN has received far less attention, despite a 2025–26 legislative turn that produced the region’s first comprehensive horizontal AI statute, a draft act, and a pre-legislative consultation paper. When ASEAN does appear, it is often portrayed as a site of norm reception or a “rule-taker” in accounts of the Brussels Effect and competition among Digital Empires. This Article provides the first comparative legal account of ASEAN’s emerging two-level order and challenges that characterization. Regionally, ASEAN’s AI governance guides establish a voluntary normative floor that converges substantially with OECD, UNESCO, UN General Assembly, and APEC frameworks. Nationally, the Article compares four contrasting approaches: Vietnam’s comprehensive statute; Thailand’s draft bill; Malaysia’s proposed framework; and Singapore’s combination of iterative soft law governance frameworks with existing general and sectoral legislation, an approach termed the “adjacent-law model”.
Using the EU AI Act as a reference point rather than a benchmark of adequacy, the comparison identifies convergence in principles but sharp divergence in scope, risk classification, institutional architecture, liability, and regulatory sandboxes. None of the four ASEAN states reproduces the EU framework. Instead, each recombines shared concepts according to domestic legal traditions, institutional arrangements, and developmental priorities. The evidence, therefore, supports selective adaptation rather than the transplantation predicted by strong accounts of a de jure Brussels Effect.
Two further findings emerge. First, the dedicated instruments are not purely regulatory: Vietnam, Thailand, and Malaysia each combine risk control with state-led capability building, treating AI legislation as a vehicle for developmental policy no less than for governance, though the relative emphasis varies across these three. Singapore pursues comparable objectives through national strategies, public investment, and ecosystem programmes outside a horizontal statute. Second, adopting a dedicated AI act is not a reliable measure of governance capacity. Singapore’s experience suggests that jurisdictions with dense general and sectoral regimes, technically capable regulators, and sufficient institutional capacity may credibly govern AI risks via existing legal infrastructure while retaining flexibility as the technology evolves.
Together, these findings reveal the “ASEAN Way” functioning largely as designed: a shared normative floor permitting tailored national architectures. ASEAN states are not merely rule-takers absorbing a dominant external model, but active participants selectively adapting shared concepts and developing distinctive designs. The region’s next challenge is therefore not legislative uniformity but regulatory interoperability among increasingly diverse national regimes, pursued as members divide across competing external alignments on AI and supply-chain security. Those alignments may also pull apart a term the governance frameworks treat as shared: trustworthiness as a property of systems, assessed through transparency and oversight, and trustworthiness as a property of origin.
Highly Recommended!
To receive new posts from Legal Theory Blog by email, get a free subscription to Legal Theory Stack.
Lawrence Solum