Official: Vietnam Launches First AI Law 2025 (No. 134/2025)

On December 10, 2025, during the 10th Session of the 15th National Assembly, the announcement became official: Vietnam launches first AI Law 2025 (No. 134/2025), marking a historic milestone in the country’s legislative history. This is the first time Vietnam has developed and promulgated a dedicated Law on Artificial Intelligence, establishing a pioneering legal framework to align with global AI development trends and bolster national competitiveness in the digital era. In the following article, Viet An Law provides a comprehensive update on the most notable new provisions within the Law on Artificial Intelligence 2025 (No. 134/2025/QH15).

Official: Vietnam launches first AI Law 2025 (No. 134/2025) – a comprehensive overview

The Law on Artificial Intelligence comprises 35 articles and 8 chapters regulating the research, development, provision, deployment, and use of artificial intelligence systems (AI activities); the rights and obligations of relevant organizations and individuals; and state management of AI activities in Vietnam. Specifically:

• Chapter 1: General provisions;
• Chapter 2: Classification and management of artificial intelligence systems based on risk;
• Chapter 3: Development of infrastructure and guarantee of national artificial intelligence sovereignty;
• Chapter 4: Application of artificial intelligence, development of innovation ecosystems and human resources;
• Chapter 5: Ethics and responsibility in artificial intelligence activities;
• Chapter 6: Inspection, examination and handling of violations;
• Chapter 7: State management of artificial intelligence;
• Chapter 8: Implementing provisions.

These regulations are clearly designed with a “management for development” approach, ensuring a balance between risk control and innovation promotion, in line with international practices and supporting Vietnam’s proactive integration with new technological standards.

The AI Law takes effect on March 1, 2026, except for the following provisions:

For AI systems that were put into operation before the effective date of the AI Law, the provider and implementer are responsible for fulfilling their compliance obligations under the AI Law within the following timeframe:

• 18 months from the effective date of the AI Law for artificial intelligence systems in the fields of healthcare, education, and finance;
• 12 months from the effective date of the AI Law for artificial intelligence systems not falling under the above provisions.

During the specified period, the artificial intelligence system may continue to operate, except in cases where the state management agency for artificial intelligence determines that the system poses a serious risk of causing damage, in which case it has the right to request a temporary suspension or termination of operation.

AI does not replace human authority and responsibility

Article 4 of the AI Law 2025 stipulates the basic principles in artificial intelligence activities as follows:

• People-centered approach; ensuring human rights, privacy rights, national interests, public interests, and national security; adherence to the Constitution and laws.
• AI serves humanity, not replaces human authority and responsibility. It ensures the maintenance of human control and the ability to intervene in all decisions and behaviors of the artificial intelligence system; system security, data security, and information confidentiality; and the ability to inspect and monitor the development and operation of the artificial intelligence system.
• Ensuring fairness, transparency, impartiality, non-discrimination, and no harm to individuals or society; adhering to Vietnamese ethical standards and cultural values; and being accountable for the decisions and consequences of artificial intelligence systems.
• Promote the development of green, inclusive, and sustainable artificial intelligence; encourage the development and application of artificial intelligence technologies that are energy-efficient, resource-saving, and reduce negative environmental impacts.

One of the core tenets behind the enactment of Vietnam’s first AI Law (Law No. 134/2025/QH15) is placing humans at the center, with the principle that AI serves humans, does not replace humans, and humans retain the right to oversight in important decisions. AI does not replace human authority and responsibility.

Prohibited AI practices: violations of data protection, IP rights, and cybersecurity regulations

Article 7 of the AI Law 2025 stipulates the following prohibited acts:

• Exploiting or hijacking artificial intelligence systems to commit illegal acts or infringe upon the legitimate rights and interests of organizations and individuals.
• Developing, providing, deploying, or using AI systems for the following purposes:
  • Committing acts that are strictly prohibited by law;
  • Using forged or simulated elements of real people or events to deceive or manipulate people’s perceptions and behavior in a deliberate and systematic manner, causing serious harm to people’s legitimate rights and interests;
  • Exploiting the weaknesses of vulnerable groups, including children, the elderly, people with disabilities, ethnic minorities, or those lacking or having limited legal capacity, or those with cognitive difficulties or impaired behavioral control, to cause harm to themselves or others;
  • Creating or disseminating false content that has the potential to seriously endanger national security, public order, and social safety.
• Collecting, processing, or using data to develop, train, test, or operate artificial intelligence systems in violation of laws on data, personal data protection, intellectual property, and cybersecurity.
• Obstructing, disabling, or falsifying human mechanisms for monitoring, intervening, and controlling artificial intelligence systems as stipulated in this Law.
• Concealing information that must be publicly disclosed, transparent, or explained; erasing or falsifying mandatory information, labels, or warnings in artificial intelligence activities.
• Exploiting research, testing, evaluation, or verification activities of artificial intelligence systems to commit acts contrary to the law.

Artificial intelligence systems are classified into three levels of risk

One of the key aspects of the first AI Law in Vietnam, No. 134/2025/QH15, is the risk-based management approach, which categorizes AI systems into different levels to ensure flexibility, encourage development, and promote technological self-reliance in AI.

Accordingly, Article 9 of the 2025 AI Law classifies the risk level of artificial intelligence systems into three levels: high, medium, and low.

• High-risk AI systems: systems that can cause significant harm to the lives, health, legal rights and interests of organizations and individuals, national interests, public interests, and national security;
• Medium-risk AI systems: systems that have the potential to confuse, influence, or manipulate users due to their inability to recognize that the interacting entity is an artificial intelligence system or the content generated by the system;
• Low-risk AI systems: systems that do not fall under the two categories above.

Accordingly, the vendor must classify the artificial intelligence system before putting it into use. Systems classified as medium risk or high risk must have a classification document attached.

However, the AI Law 2025 currently lacks clear quantitative or qualitative criteria, regulations on management measures, and assessment tools, leading to confusion for enterprises in self-classifying risks and concerns about legal liability. These issues will be addressed in a Government Decree in the near future.

High-risk AI systems: mandatory conformity assessment prior to deployment

According to Clause 1, Article 13, high-risk AI systems must undergo conformity assessment before being put into use or when significant changes occur during use. If there are standards or technical regulations for artificial intelligence systems, then conformity assessment must also be carried out in accordance with the law on standards and technical regulations.

Accordingly, the assessment is carried out by a conformity assessment body that is registered or recognized in accordance with the law. The Prime Minister stipulates the List of high-risk AI systems, including the list of artificial intelligence systems that must be certified for conformity before being put into use.

Many policies support enterprises operating in the field of artificial intelligence

According to Clause 1, Article 20 of the AI Law 2025, organizations and individuals operating in the field of artificial intelligence are entitled to the highest incentives and support as prescribed by the law on science and technology, investment, digital technology industry, high technology, digital transformation, and related laws; they are provided with conditions to access infrastructure, data, and testing environments serving research, production, and commercialization of artificial intelligence products and services.

In addition, small and medium-sized enterprises and innovative startups in AI also benefit from various support policies:

• Priority access to technical infrastructure, data, and testing environments; support in terms of costs, training, and market connections for the development of AI products and services (Clause 4, Article 20);
• Supported with conformity assessment costs; provided free of charge with sample documents, self-assessment tools, training, and consulting; given priority for support from the National Artificial Intelligence Development Fund (Clause 1, Article 25);
• Supported through support vouchers to use computing infrastructure, shared data, large-scale Vietnamese and ethnic minority language models, training platforms, testing, and technical consulting services for research, development, and deployment of AI applications (Clause 2, Article 25);…

This regulation creates a comprehensive legal framework and support mechanism to encourage organizations and individuals, especially small and medium-sized enterprises and innovative startups, to promote research, development, and commercialization of AI products and services, contributing to the promotion of national innovation and digital transformation.

Regulations establishing a controlled testing (sandbox) mechanism for artificial intelligence

According to Article 21 of the AI Law 2025, the controlled testing mechanism for artificial intelligence shall be implemented in accordance with the law on science, technology, and innovation. The results of controlled testing shall serve as the basis for consideration by competent state agencies.

• Recognition of conformity assessment results;
• Exemption, reduction, or adjustment of corresponding compliance obligations.

The primary purpose of a controlled testing mechanism for AI is to remove legal barriers hindering AI models, allowing for limited-scale testing to gather data demonstrating effectiveness, thereby fostering long-term regulatory reform.

AI system liability: Compensation responsibilities of the implementing party

According to Clause 2, Article 29 of the AI Law 2025, if a high-risk AI system is managed, operated, and used in accordance with regulations but still results in damage, the implementing party shall be responsible for compensating the injured party. After compensation, the implementing party may request the supplier, developer, or related parties to reimburse the compensation amount if there is an agreement between the parties.

This liability for damages is exempted in the following cases:

• The damage occurred entirely due to the intentional fault of the injured party;
• The damage occurred in a case of force majeure or emergency, unless otherwise provided by law.

Thus, to enhance the protection of citizens, the Law has stipulated a strict mechanism for compensation for damages, specifying the responsibility of relevant parties to compensate for damages even if they are not at fault.

It can be affirmed that the enactment of the AI Law 2025 No. 134/2025/QH15, the first of its kind in Vietnam, is a turning point, creating a pioneering legal framework to help Vietnam catch up with the global AI development trend and enhance national competitiveness in the digital age.

It can be affirmed that the milestone, “Official: Vietnam launches first AI Law 2025 (No. 134/2025)”, represents a landmark achievement, establishing a pioneering legal corridor for Vietnam to align with global AI development trends and enhance national competitiveness in the digital era.

If you have any questions or require legal advice regarding AI regulations, please contact Viet An Law for the best consultation and support!