Towards AI Policy for the Asia-Pacific

By Mai Nguyen

Dr Nhat-Mai Nguyen (Mai Nguyen), Deputy Director of the Hub for Vietnam Policy Studies at the Australian National University, examines the emerging policy challenges of artificial intelligence (AI) across the Asia-Pacific. She argues that moving from AI strategy to risk-based, human-centred governance is critical to harnessing AI for public value, while highlighting the opportunities and challenges for Vietnam.

AI Has Moved Beyond Technology Policy ‍

AI is rapidly becoming one of the most consequential public policy issues of the twenty-first century. Unlike previous waves of digital transformation, AI is advancing faster than many governments can understand, govern, and regulate it. Increasingly powerful foundation models now demonstrate capabilities in reasoning, programming, scientific discovery, content creation, and autonomous decision-making. Yet legislative and policymaking processes continue to operate at a much slower pace.

This widening gap between technological innovation and governance capacity is becoming a strategic risk. If policy responses are too slow, AI may become deeply embedded across economies and societies before adequate mechanisms for safety, accountability, and protection of the public interest are established.

The Asia-Pacific is at the centre of this transformation. The region includes leading technology hubs, advanced industrial economies, and countries undergoing rapid digital transformation. AI offers major opportunities to raise productivity, improve public services, accelerate scientific research, and transform education and healthcare. At the same time, it raises difficult questions about employment, privacy, cybersecurity, intellectual property, the distribution of benefits, and the responsibilities of governments and technology companies.

The question, therefore, is no longer whether countries need AI policies. It is whether they can build governance systems capable of keeping pace with technological change while protecting people and the broader public interest. The next few years represent an important policy window for getting that balance right.

AI Is No Longer Just a Technology Story

For much of the past decade, governments approached AI primarily as an innovation and competitiveness issue. Policy focused on supporting research, attracting investment, developing digital infrastructure, and encouraging business adoption. AI was treated as an enabling technology capable of improving productivity and creating new industries.

That perspective is no longer sufficient.

Frontier AI systems increasingly perform tasks once considered the domain of skilled professionals: analysing complex information, writing software, generating scientific hypotheses, creating content, and supporting decision-making. AI agents are also becoming capable of planning and executing sequences of tasks with increasingly limited human intervention.

More importantly, these capabilities are developing remarkably quickly. By the time regulators begin to understand one generation of models, another may already be deployed with substantially greater capabilities. AI has consequently moved beyond technology policy to become an issue of economic policy, labour markets, law, security, public administration, and international cooperation.

The same technology may generate substantial productivity gains while disrupting cognitive work; accelerate drug discovery while increasing the potential misuse of biological knowledge; and improve public administration while expanding the capacity to analyse personal behaviour.

Voluntary commitments to transparency and safety were valuable during the earlier stages of AI development. But as frontier models become more capable and potentially create serious cybersecurity, biological, autonomous-system, or critical-infrastructure risks, voluntary mechanisms alone are unlikely to be sufficient.

This does not mean regulating every AI system in the same way. Governance should be risk-based, concentrating the strongest requirements on highly capable models and high-impact applications. The objective should be to reduce serious risks without unnecessarily obstructing socially beneficial innovation.

The Real Challenge Is Governance

The consequences of transformative technologies are shaped not only by their technical capabilities but by the institutions and rules surrounding them. For AI, the central challenge is therefore the capacity of governments to govern a technology that is evolving rapidly and whose most advanced capabilities are concentrated among relatively few actors.

Regulation of aviation and pharmaceuticals offers a useful analogy. Aircraft cannot enter commercial operation without safety assessment, and medicines must undergo evaluation before reaching patients. Similarly, frontier AI systems capable of creating systemic risks should be independently evaluated before large-scale deployment.

A proportionate framework could include mandatory safety testing, cybersecurity requirements for developers, incident reporting, regular auditing, and regulatory authority to intervene when systems present serious risks. The purpose is not to impose identical obligations on every AI application, but to match oversight to potential harm.

Stronger governance should not, however, mean slowing innovation across the board. AI has enormous potential in science and medicine, including drug discovery, protein design, medical imaging, personalised treatment, and research analysis. While powerful AI models may require stronger oversight, regulatory processes for beneficial AI-enabled scientific and medical innovations should become more agile and responsive.

Labour markets require similar attention. Unlike earlier automation, which predominantly affected manual and repetitive tasks, generative AI can reshape administrative, professional, and creative work. It may generate significant economic growth, but its benefits and transition costs will not necessarily be evenly distributed. Governments therefore need better measurement of AI's effects on jobs, skills, and income; stronger retraining and lifelong learning systems; support for occupational transitions; and incentives for employers to redeploy workers where possible. Longer-term income support may also need consideration if structural displacement becomes widespread.

Governance must also address privacy, personal data, and high-impact automated decisions. People should be able to understand and challenge consequential decisions involving AI. Governments and technology companies alike therefore need appropriate oversight, transparency, and accountability for how powerful systems are deployed.

From AI Strategy to Public Value in the Asia-Pacific

For the Asia-Pacific, success should not be measured simply by how quickly AI is adopted. The more important question is whether AI creates lasting public value.

First, international cooperation should become a central element of AI policy. AI models are developed, hosted, and provided across borders, making purely national governance increasingly inadequate. Countries need greater cooperation on safety standards, evaluation mechanisms, incident information, semiconductor supply chains, cybersecurity, and trusted AI technologies.

Second, AI policy should support broader economic and social objectives rather than focusing narrowly on investment, data centres, or the number of AI companies. Productivity and growth matter, but so do public-service quality, access to technology, skills development, and the digital divide.

Third, governments need AI governance capacity within the public sector itself. Public officials must understand both the capabilities and limitations of AI when it is used in policymaking or service delivery. Public procurement should therefore incorporate requirements for data quality, security, auditability, human oversight, and vendor accountability.

Intellectual property illustrates the same need for balance. Debate in Australia over copyrighted works and AI training has highlighted tensions between encouraging AI development and protecting creators. Rather than treating these objectives as incompatible, transparent licensing arrangements between AI developers and rights holders can support innovation while preserving consent and reasonable compensation.

The broader principle is straightforward: governments should neither obstruct useful innovation nor assume that technological progress automatically produces public value. Policy must establish boundaries, allocate responsibility, and ensure that the benefits of AI are shared more widely.

AI Governance in Vietnam: Opportunities and Challenges

Vietnam provides an important example of how these regional challenges are becoming national policy questions. As an open, dynamic economy with a young population and an ambitious digital transformation agenda, Vietnam has opportunities to use AI to raise productivity and improve public governance. Applications in manufacturing, logistics, finance, agriculture, education, healthcare, and public services could contribute to economic transformation.

In government, AI can support data analysis, policy forecasting, fraud detection, administrative simplification, and more personalised services. In healthcare, it can assist diagnosis and improve access where specialist personnel are limited. In education, AI can support teachers and help tailor learning to individual needs. Vietnamese businesses, including small and medium-sized enterprises, can also use AI to automate processes, analyse markets, and participate more deeply in regional and global value chains. ‍

But these opportunities come with significant constraints. Data remain fragmented across many organisations and vary considerably in quality. Poor or biased data can produce unreliable AI outputs and weaken decision-making. Vietnam also faces limitations in advanced computing capacity, specialised human capital, and access to frontier technologies. Without careful policy, AI could widen gaps between large and small businesses and between urban and rural areas.

The regulatory framework must therefore evolve alongside adoption. Vietnam needs continued attention to personal data, cybersecurity, accountability for AI-related harm, algorithmic transparency, and high-impact applications. Crucially, regulation should distinguish low-risk uses from systems that directly affect people's income, health, opportunities, or other important interests.

Vietnam also needs to move from AI strategy towards implementation and oversight capacity. This requires stronger data and computing infrastructure, specialised talent, AI capability among public officials, controlled experimentation, and practical guidance for government use.

Priority should be given to areas where AI can generate clear social value, including primary healthcare, education, agriculture, disaster management, and administrative reform. Success should be judged not by technological sophistication alone, but by measurable improvements in productivity, service quality, and people's lives.

A central principle should be that AI supports human decision-making rather than replacing human responsibility. For high-impact decisions, institutions and officials must retain ultimate accountability. If Vietnam can develop a flexible, risk-based, human-centred governance framework, it can learn from international experience while adapting AI governance to its own development needs.

Conclusion

AI has entered a phase in which governance capacity may matter as much as technological capability. Its development is moving faster than traditional policymaking, giving governments a limited window to establish appropriate institutions before AI becomes deeply embedded throughout economic and social life.

A balanced policy agenda must therefore pursue several objectives simultaneously: risk-based oversight of powerful models, support for workers facing economic transition, more agile regulation of beneficial scientific innovation, protection of privacy and personal data, and stronger regional and international cooperation.

For Vietnam and the wider Asia-Pacific, AI policy should ultimately be about more than keeping pace with technology. Its purpose should be to direct technological progress towards economic and social development, stronger public governance, and tangible benefits for people.

AI can become an important engine of growth and innovation, but that outcome is not automatic. It will depend on choices about institutions, investment, skills, data, and accountability made today. Countries that encourage innovation while managing its risks—and judge AI by the public value it creates—will be better positioned to shape an AI future that is safe, inclusive, and sustainable.