In September 2025, the Government of Vietnam released the Draft Law on Artificial Intelligence (“Draft AI Law”) for public consultation. Scheduled to be enacted during the National Assembly’s 10th session starting on 20 October 2025, the Draft AI Law will create Vietnam’s first comprehensive legal framework for the development, deployment, and utilization of artificial intelligence.
The Draft AI Law establishes foundational principles for AI development, delineates the roles and responsibilities of government authorities and AI developers, and introduces new schemes to regulate, manage, and promote AI innovation nationwide.
Below are the key highlights of the Draft AI Law:
1. Broad Scope
The Draft AI Law asserts broad jurisdiction, applying not only to domestic AI activities but also holding international entities accountable if their AI systems impact the Vietnamese market, users, or legal interests
It explicitly excludes AI systems that are exclusively developed, deployed, or used for national defence, security, or intelligence purposes—requiring entities in these sensitive sectors to maintain separate compliance frameworks.
2. Fundamental Definitions for AI
Article 3 of the Draft AI Law sets out a series of fundamental definitions, though several remain conceptually broad and open to interpretation.
For instance, “AI” is defined as a branch of computer science focused on creating machine-based systems capable of performing tasks requiring human intelligence. However, the definition of an “AI system” omits an explicit reference to AI, describing instead an autonomous machine-based system influencing physical or digital environments.
Similarly, terms such as “National AI Infrastructure” and “National AI Autonomy Capacity” are broadly drafted, emphasizing independence and integration without defining clear technical parameters. This signals a likely need for further government guidance to ensure consistent interpretation and practical application across industries.
Businesses should therefore not assume that only companies with “AI” in their branding will be affected. The definitions in the Draft AI Law could capture a broad spectrum of tech companies, including those in IoT, advanced analytics, and process automation.
3. Risk-Based Classification of AI Systems
Article 8 of the Draft AI Law introduces a tiered risk-based regulatory framework, classifying AI systems into four categories:
- Unacceptable risk – Prohibited systems posing threats to national security or human rights.
- High risk – Systems used in critical sectors such as healthcare, finance, or transportation, requiring prior government approval.
- Medium risk – Systems involving direct human interaction or content generation, subject to transparency requirements.
- Low risk – All other systems, subject to minimal oversight.
Before launching any AI product in Vietnam, companies must conduct a self-assessment and maintain documentation. For systems considered high-risk, obtaining prior government approval is mandatory. Furthermore, all foreign providers are required to appoint a legal representative in Vietnam to handle compliance and act as the main contact for authorities.
4. Prohibited AI Applications
Article 11 of the Draft AI Law prohibits AI applications deemed to pose unacceptable risks, including:
- Manipulating human cognition or behavior to cause physical or psychological harm.
- Social credit scoring that leads to unfair discrimination.
- Real-time remote biometric identification in public spaces for law enforcement (except in narrowly approved serious crime cases).
- Development of large-scale facial recognition databases from indiscriminate data scraping.
- Creation or dissemination of deepfakes or AI-generated content that threaten national security, public order, or social stability.
Violations may result in administrative sanctions, criminal prosecution, or civil liability, depending on severity.
5. Registration with the National AI Database
Article 22 of the Draft AI Law establishes a National AI Database to enhance transparency and accountability.
High-risk AI systems must be registered before market entry or deployment. Entities contributing data to the database may enjoy preferential treatment, such as tax incentives, financial support, and priority access to national AI infrastructure.
6. Public–Private Partnerships for National AI Infrastructure
Article 31 of the Draft AI Law introduces a groundbreaking mechanism allowing data and AI models to be treated as lawful capital contributions in public–private partnership projects.
This represents a further step towards Vietnam’s legal recognition of intangible digital assets, empowering private investors to contribute technological resources alongside traditional capital to accelerate the country’s AI infrastructure development.
7. National AI Development Fund and Support Schemes for Start-ups and SMEs
Article 38 of the Draft AI Law establishes the National AI Development Fund, a non-budgetary financing vehicle supporting AI research, innovation, and capacity building.
Funded through state allocations and private contributions, the Fund will provide grants and loans to strategic AI projects and prioritize startups and SMEs.
Article 49 of the Draft AI Law complements this initiative by mandating targeted financial and regulatory support for smaller enterprises, including tax incentives, loans, R&D vouchers, and participation in AI regulatory sandboxes. Training and streamlined administrative procedures further aim to integrate SMEs into Vietnam’s evolving AI ecosystem.
8. Establishment of AI Clusters
Under Article 50 of the Draft AI Law, the Government may establish AI Clusters—dedicated innovation zones concentrating talent, data resources, and infrastructure.
Enterprises operating within these clusters will benefit from preferential policies, such as access to government-funded research facilities, shared data pools, tax reductions, land-use incentives, and fast-tracked administrative procedures. Collaboration between universities, research institutes, and private enterprises will be actively encouraged to drive AI commercialization.
9. Government Intervention and Remedies
Finally, Article 57 of the Draft AI Law empowers authorities to intervene in AI operations that violate regulations or cause harm. Measures may include orders to suspend, restrict or terminate AI activities.
Affected providers will be notified and entitled to appeal or initiate legal proceedings if they believe the intervention to be unjustified. These provisions seek to balance public safety and private rights, ensuring proportional enforcement.
Key Takeaways for AI Investment
Vietnam’s upcoming Draft AI Law provides a predictable roadmap for the future, with a phased rollout beginning January 1, 2026. This staggered timeline gives businesses a strategic window to align with the new standards, starting with key obligations in 2027.
This forward-thinking framework signals Vietnam’s ambition to build a world-class, innovation-driven AI ecosystem. By balancing rapid technological advancement with safety and ethics, the government is creating a stable and trusted environment where cutting-edge AI can thrive responsibly.
The message to AI pioneers is clear: now is the time to lead. By proactively assessing your AI systems, fortifying your data and security protocols, and investing in your team’s capabilities, you can not only ensure seamless compliance but also position your business at the forefront of Vietnam’s exciting AI revolution.
Should you require any assistance navigating the Draft AI Law or its potential implications for your business, please feel free to contact DFDL.
The information provided here is for information purposes only and is not intended to constitute legal advice. Legal advice should be obtained from qualified legal counsel for all specific situations.
