Vietnam AI Law Explained: What the New Rules Mean for AI Development and Deployment

Vietnam has entered a new phase of AI governance. As of March 1, 2026, the Vietnam AI Law is officially in force, establishing the country’s first comprehensive framework for regulating artificial intelligence activities.

Passed by the National Assembly on December 10, 2025, the law regulates the research, development, provision, deployment, and use of AI systems in Vietnam. It introduces a risk-based regulatory approach, establishes obligations for high-risk AI systems, and clarifies accountability across the AI lifecycle from developers and providers to deployers and end users.

For organizations operating AI systems in Vietnam, the law provides an early view of how the country intends to balance innovation with safeguards around safety, transparency, and responsibility.

A Risk-based Framework for AI Governance

At the center of the Vietnam AI Law is a risk-based governance model.

AI systems are categorized by their level of risk (low, medium, or high) with stricter obligations applying to systems that may have greater impact on individuals, public safety, or national infrastructure.

High-risk AI systems face the most stringent requirements. Before deployment, these systems must undergo conformity assessments to verify that they meet safety and compliance requirements. Depending on the category, assessments may be conducted either through registered conformity assessment bodies or through self-assessment mechanisms.

This model allows lower-risk AI applications to develop with greater flexibility while ensuring that systems capable of causing harm receive closer oversight.

In practice, this means organizations must first evaluate the risk level of their AI systems before deploying them. For example, a generative AI tool used internally to draft marketing copy may fall into a lower-risk category. By contrast, an AI model used to evaluate credit eligibility or recommend medical treatment would likely attract stricter scrutiny due to its potential impact on individuals.

Clarifying Roles Across the AI Lifecycle

Another core feature of the Vietnam AI Law is the clear definition of roles involved in AI development and deployment.

The law distinguishes several actors:

• Developers, who design, build, train, or fine-tune AI models or systems
• Providers, who place AI systems on the market under their name or brand
• Deployers, who use AI systems in professional or commercial contexts
• Users, who interact directly with AI systems or rely on their outputs
• Affected persons, whose rights or interests may be impacted by AI outcomes

This structure reflects the reality that modern AI systems often involve multiple parties across the technology supply chain.

For example, a global software provider may develop an AI-powered hiring tool, while a Vietnamese company deploys the tool to screen job applicants. Under the law, both parties may carry responsibilities depending on their role in designing, providing, or operating the system.

By assigning responsibilities across the AI lifecycle, the Vietnam AI Law aims to ensure that accountability does not disappear once a system moves from development to real-world use.

Additional Safeguards for High-risk AI Systems

Organizations operating high-risk or significantly impactful AI systems must meet several additional obligations.

These include conducting impact assessments where AI systems are used in state management or public service provision. These assessments evaluate risks associated with the system and identify measures to mitigate those risks. They must address issues such as safety, human oversight, and safeguards against harm, although certain information may remain confidential where necessary.

More broadly, high-risk AI systems must meet general requirements around risk management, technical documentation and logging, transparency, and incident handling.

Entities deploying these systems must also maintain ongoing compliance and provide disclosures about how the systems operate.

In practical terms, this may involve documenting how the system processes input data, what decisions or recommendations it generates, and what oversight mechanisms exist to ensure those outputs are used responsibly.

For instance, a financial institution deploying an AI-driven credit scoring system may need to assess how the model affects individuals’ access to services, identify potential bias or inaccuracies, and establish review processes where humans can intervene when necessary.

Prohibited Uses of AI

The Vietnam AI Law also establishes clear prohibitions on certain uses of AI.

These include deploying AI systems for unlawful purposes or for manipulative or deceptive practices that cause serious harm. The law also explicitly prohibits AI uses that exploit vulnerable groups or generate deceptive content designed to defraud individuals or harm their reputation.

Examples of prohibited practices may include:

• Using AI systems to manipulate user behavior or perception in ways that cause harm
• Generating synthetic content intended to deceive or defraud individuals
• Exploiting children, older adults, or people with disabilities through AI-enabled systems
• Deploying AI systems in ways that threaten national security or public order

These prohibitions reflect a growing global focus on limiting harmful uses of AI while still allowing innovation in lower-risk applications.

Enforcement and Accountability

The law introduces a range of enforcement mechanisms to address violations.

Organizations that fail to comply with the Vietnam AI Law may face administrative sanctions, criminal liability, or civil compensation obligations where damages occur.

Importantly, the law also addresses situations where harm arises even when a high-risk AI system has been used in a compliant manner. In these cases, the deployer of the system remains responsible for compensating affected individuals, although the deployer may seek contractual recovery from the provider or developer responsible for the system.

Serious incidents must be reported through the national AI one-stop portal, which enables authorities to investigate and take corrective action. Depending on the circumstances, regulators may order the suspension or recall of AI systems, and enforcement outcomes may be made public.

The framework also introduces the concept of serious incidents, defined as events linked to AI system operation that cause (or risk causing) significant harm to individuals, human rights, property, cybersecurity, public order, or national security systems.

This definition signals that authorities expect organizations to monitor the real-world impact of their AI systems and respond when incidents occur.

A National Strategy for AI Development

Beyond compliance obligations, the law also signals Vietnam’s broader ambitions for AI development.

It encourages investment in national AI infrastructure, including computing capacity, data sharing initiatives, and the development of foundational AI models. The framework also supports public-private cooperation to accelerate innovation while maintaining appropriate safeguards.

At the same time, the law emphasizes a consistent principle: AI should support human decision-making rather than replace it in matters of significant societal importance.

Organizations developing or deploying AI systems in Vietnam should now evaluate how their technologies fit within the law’s risk categories, assess whether additional obligations apply to their systems, and ensure governance processes are in place to manage compliance.

For a deeper look at the Vietnam AI Law, explore the detailed regulatory analysis available in OneTrust DataGuidance.

You can also explore additional resources covering Vietnam’s privacy and technology regulations, including broader data governance and digital policy developments across the region.

To operationalize AI governance and manage regulatory requirements across jurisdictions, organizations can also explore OneTrust AI Governance solutions, designed to help map AI systems, manage risk assessments, and support responsible AI oversight.

Key Questions About the Vietnam AI Law