On March 1 2026, Vietnam’s AI Law officially took effect, after being passed during the National Assembly’s 10th session in December 2025. From that, Vietnam has become the first country in Southeast Asia, and among a handful of countries globally, alongside the EU, South Korea and Japan to establish a legal framework for the use of artificial intelligence.
This new law aims to promote artificial intelligence as “a key driver of growth, innovation and sustainable development.” Key features include labelling AI-generated content, evaluating AI risk, and introducing a new AI information centre and database.
A law aimed to encourage AI growth
As AI advances rapidly, bringing both significant benefits and risks, the debate of how to regulate this technology has been raised among policymakers.
In practice, only a handful of countries have introduced formal legal frameworks for the use of AI, including South Korea, Japan and the European Union. Most other countries prioritise a softer approach, which focuses on voluntary guidelines and ethical standards.
As Vietnam moves toward the “Era of Rising,” the application of artificial intelligence is becoming one of the national goals. The governments have stated the new law is to promote AI as “a key driver of growth, innovation and sustainable development.”
The law, therefore, seeks not only to manage the risks associated with AI but also to encourage AI adoption across key sectors such as healthcare and education.
It also states that the fundamental principle of AI is to serve humans, rather than replace human authority and responsibility. Humans retain oversight and the ability to intervene in the decisions and actions of AI systems.
Four key features of Vietnam’s AI Law
The Artificial Intelligence Law (No. 134/2025/QH15), which officially took effect on March 1, 2026, consists of eight chapters and 35 articles. Several key provisions include:
Labelling AI content
Article 11 sets out clear transparency requirements for the development and use of AI. Under the provision, all audio, images and videos generated by AI must carry machine-readable labels to distinguish them from authentic content. The measure is seen as an important step toward preventing the spread of deepfakes. The law does not yet specify detailed penalties or enforcement measures.
AI risk evaluation system
Under Article 9, AI systems are classified into three risk levels, each subject to different regulatory requirements.
- High-risk AI refers to systems that could potentially cause significant harm to human life, health, human rights, national security or the public interest. They are subject to the strictest monitoring.
- Medium-risk AI includes systems that may mislead or manipulate users, particularly when individuals are unaware that they are interacting with artificial intelligence, and could influence users’ behaviour
- Low-risk AI refers to systems that do not fall into the two categories above, typically simple support tools or applications with limited impact.
Highest incentives for AI development
The law leaves one whole section to present incentives and support to organisations and individuals developing core AI technologies. Under Article 20, AI companies are eligible for the highest level of incentives available under laws governing science and technology, high-tech industries, digital transformation and investment.
Integrating AI into the school curriculum
One provision of the law is to teach basic AI knowledge, computational thinking, digital skills and technology ethics in the mandatory education curriculum. Meanwhile, vocational institutions and universities are also encouraged to develop training programs in artificial intelligence, data science and related fields.
The potential trade-offs of early AI regulation
In January 2026, South Korea officially brought its AI Basic Act into force, making it the first comprehensive AI law in the world to take effect. With this move, Seoul is positioning itself at the forefront of shaping global rules for artificial intelligence, as part of its ambition to become one of the world’s top three AI powers.
However, as the law functions largely as a framework and lacks detailed implementation guidelines, companies, particularly startups, may find it difficult to determine the boundaries of compliance. Penalties of up to 30 million won (about US$20,400) for not labelling AI-generated content also pose legal risks that could discourage smaller startups with limited resources from pursuing more aggressive technological innovation.
In comparison, Vietnam’s AI Law takes a more cautious approach, seeking to balance regulation with room for technological development. The legislation does not specify detailed penalties, leaving the specifics to be outlined in future decrees and implementing guidelines.
As a framework law, the legislation does not immediately change how AI is used in Vietnam. Instead, it establishes the legal basis for future regulations of artificial intelligence and sets an example for similar regulations in other countries
However, South Korea’s experience highlights the potential trade-offs of early AI regulation. On one hand, it positions Vietnam among the pioneering countries seeking to regulate artificial intelligence. On the other hand, any misstep could expose the country to unprecedented challenges as the technology continues to evolve rapidly.
The new AI law and the regulations that follow, therefore, may test Vietnam’s ability to navigate technologies in this new stage of development.
