The National Assembly adopted the Law on Artificial Intelligence No. 134/2025/QH15 (“AI Law”) with effect from March 1, 2026. Vietnam has now established a comprehensive regulatory framework for Artificial Intelligence (“AI”) technology. At its philosophical core, the law adopts a human-centric approach–echoing the spirit of the European Union’s AI Act–while carving out distinctly Vietnamese structural features. Most notably, it creates a dedicated national AI ethical framework.
A Foundation Built on Human-Centric Principles
By requiring that all AI activities be placed with “man at the center”, Vietnam places human rights protection at its core (Article 4 of the AI Law). Protection of human rights and privacy are treated as non-negotiable priorities. Crucially, AI is seen to serve, not to replace, human responsibility. Critically, high-risk systems must be designed to allow for effective human oversight–giving natural persons the ability to intervene, override, or shut down a system that is not performing as intended (Article 14 of the AI Law).
Article 7 sets out a firm list of prohibited practices. The law bans the use of AI to forge images or sounds, or to deploy deceptive techniques that manipulate human perception for harmful purposes. It also forbids AI applications that exploit the vulnerabilities of protected groups, including children, the elderly, persons with disabilities and ethnic minorities, in ways that cause self-harm or harm to others. The creation or dissemination of content that seriously threatens national security, social order, or public safety is equally prohibited.
A National AI Ethical Framework
Article 26 introduces an affirmative National AI Ethical Framework. This framework is grounded in principles that AI must not harm the life, health, honor, or dignity of individuals; must respect human rights and ensure fairness and non-discrimination; must actively promote the happiness and sustainable development of society as a whole; and must encourage innovation and social responsibility in the research, development, and application of AI.
For AI that is deployed in public management or for public services, the law requires that high-risk systems or systems that have a significant impact must make documented assessments of the impact on human rights, fairness, and collective social interests (according to Article 27).
On March 10, 2026, a few days after the effective date of the AI Law, the Ministry of Science and Technology issued Circular No. 05/2026/TT-BKHCN (“Circular 05”) with immediate effect. Circular 05 gives operational details of the National AI Ethical Framework that applies directly to state agencies, organizations, and individuals involved as developer, supplier, implementer, or user of AI systems to serve public administration or to provide public services. Vietnamese and foreign organizations and individuals who are active in research, development, supply, implementation, or who use AI systems in Vietnam are encouraged to follow it (Article 1 of Circular 05).
Under this framework, AI ethics is defined as a system of values, principles, and standards that guide the research, development, provision, deployment, and use of AI to ensure respect for human beings, human rights, the public interest, and sustainable development. The framework is to be reviewed and updated every three years, or whenever significant changes in technology, law, or management practice occur.
Circular 05 reinforces human rights protection by presenting several concrete principles. It requires that organizations and individuals build monitoring and intervention mechanisms that ensure that humans retain meaningful control over every AI decision–and that in public services, final decision-making authority always rests with a human being, and is never delegated to an AI system. It also mandates proactive review measures to ensure that AI systems do not infringe upon personal data, privacy, free will, or the right to information. On the question of fairness, Circular 05 requires organizations and individuals to use measures to identify and mitigate data bias, model bias, and operational bias, and again, with particular attention to the impact of AI on vulnerable groups including children, the elderly, people with disabilities, and disadvantaged groups. To close the accountability loop, systems must be designed to provide explainable, evidence-backed reasoning for their decisions, and the party responsible for those decisions must be clearly identified.
Protecting Data Privacy
Data privacy is not treated as a subsidiary concern. It is a core principle throughout the AI Law. Article 4 establishes the “right to privacy” (quyền riêng tư) as a foundational value, meaning that every stage of the AI lifecycle, from research to deployment, must prioritize the protection of individual information. Article 7 of the AI Law reinforces this by explicitly prohibiting the illegal collection, processing, or use of data to develop, train, test or operate AI systems in a way that violates existing personal data protection and cybersecurity laws.
For sensitive sectors–healthcare and education–Article 6 of the AI Law strictly mandates heightened data protection standards, requiring that AI applications in these fields actively safeguard the personal data of patients and learners. Articles 16, 17 and 31 of the AI Law introduces the development of the National AI Infrastructure[1]. The National AI Infrastructure must comply with laws on personal data protection and intellectual property. State agencies receiving organizational and personal data for AI management are legally bound to preserve the safety, security, and confidentiality of that data, including business secrets and personal information.
Circular 05 builds on these foundations by integrating privacy protection into its core principles of safety and reliability. Stakeholders are required to apply appropriate measures to detect, prevent, and respond to data leaks, data poisoning, model poisoning and unauthorized access, etc, and to ensure the confidentiality, integrity, and availability of all data, models, algorithms, and relevant infrastructures. Developers, providers and implementers are specifically required to implement access control, data traceability, and lifecycle data governance to protect personal and private information. They must also clearly define the data protection and ethical compliance obligations of all parties, including third-party contractors and partners. Users, too, are instructed to take responsibility for their own information and business secrets, refraining from sharing sensitive personal with AI systems without authorization or safeguards (Appendix I of Circular 05).
Enforcement and Accountability
To promote enforcement, the AI Law establishes a clear penalty regime. Under Article 29, organizations and individuals who violate the AI Law or related laws in areas such as data protection and cybersecurity, face administrative fines or criminal prosecution, depending on the severity and consequences of the violation. Civil liability for damages is also a remedy.
Circular 05 supports compliance through several practical tools. Appendix II provides a Sample Self-Assessment Form that allows organizations to evaluate their own adherence to principles including safety, human rights, and social well-being. Stakeholders are required to establish dedicated channels to receive and handle complaints related to bias, discrimination, harmful content, or system misuse. Organizations are encouraged to develop their own internal codes of AI ethics, tailored to the specifics of their industry, to complement the national framework.
Conclusion
Vietnamese AI Law and the National AI Ethical Framework under Circular 05 represent a significant step toward responsible AI governance. By anchoring regulation in human-centric principles, setting clear prohibitions against harmful and manipulative uses of AI, mandating robust data privacy protections, and creating a pathway for ethical compliance, Vietnam has signaled that technological progress and the protection of fundamental rights are inseparable.
[1] National AI Infrastructure is a strategic infrastructure that includes infrastructure invested in by the State, enterprises and social organizations. It is developed as a uniform, open and secure ecosystem that is capable of connectivity, sharing and expansion in accordance with requirements for development and application of AI.
