1.

Jurisdiction
 
The Vietnamese AI Law applies to both domestic and foreign organizations and individuals “participating in AI activities in Vietnam”. Such term is intended to encompass the full lifecycle and application of AI systems, from R&D to provision, deployment and use of AI. In comparison, the jurisdiction of the Korean AI Act covers conducts that take place outside of Korea as long as they affect domestic market or users in Korea.
 

2.

Definitions of AI Participants
 
Under the Vietnamese AI Law, a “developer” is an organization or individual who (i) engages in designing, developing, training, testing or adjusting all or part of a model, algorithm or system of AI; and (ii) has the direct control over the technical methods, training data or model parameters. On the other hand, a “provider” is an organization and individual who provides AI system to the market, or puts it into operation under their own name, brand or trademark regardless of whether the system is developed by themselves or by a third party.
 
The Vietnamese AI Law further defines “deployer” as who uses AI system under its control in professional, commercial and service activities, and “user” as who directly interacts with the AI system or uses its output.^[1] These definitions closely align with the EU AI Act.
 
Compared to the Korean AI Act, which combines developers, providers and deployers into one category as “AI business operators”, the Vietnamese AI Law follows EU model more closely, but distinguishing developers and deployers from other AI participants by adding a “control” requirement. In other words, organizations that fall within the scope of “AI business operators” under the Korean AI Act, but who do not have control of the technology, could possibly be not considered as “developers” or “deployers” under the Vietnamese AI Law.
 

3.

Risk-based Classification
 
Under the Vietnamese AI Law, AI systems are categorized according to a three-tiered risk-based classification, which is a major difference as compared to the Korean AI Act.^[2] The providers bear the responsibility of classifying AI systems according to the current criteria specified in Decree 142.
 

Under Decree 142, AI system providers must conduct re-evaluation and re-classification of AI system’s risk level under the following circumstances: (i) the AI system undergoes significant changes affecting the initial risk classification criteria or it experiences a serious incident demonstrating actual higher risk level, or (ii) there is a decision from the Prime Minister or a written request from competent state agency to change the current AI system’s risk level.^[4]
 

4.

Transparency Requirements
 
The transparency requirements under Vietnamese AI Law share the same goals with the Korean AI Act, namely, preventing deception and promoting user awareness with respect to AI. Under Vietnamese AI Law, specific role-based transparency requirements are as follows:
 

• Providers must ensure that (i) AI systems interacting directly with humans are designed and operated so that users can recognize that they are interacting with AI and (ii) AI-generated contents are marked in a machine-readable format as prescribed by the government;^[6] and
   

• Deployers must (i) make clear notifications when their provided AI-generated contents may cause confusion and (ii) clearly label to distinguish AI-generated contents that mimic human appearance, voice or actual events from human-made contents.^[7]
   

Meanwhile, the Korean AI Act and its guiding Presidential Decree place transparency burden with respect to “High-Impact AI” and “Generative AI” on all business operators, who must notify users in advance through methods such as labelling, providing cautionary statement or informational posting, as prescribed by the government.^[8]
 
Notably, both the Vietnamese AI Law and Korean AI Act include relaxed requirements for artistic and creative works, such as the labeling can be done in a manner as to not hinder the presentation or enjoyment of such works. 
 

5.

Serious Incident Management
 
Serious incidents causing harms to life, property or national security or causing disruptions to public and necessity services must be reported via the Single-Window Website on AI^[9] managed by the Ministry of Science and Technology. Specifically, for serious incidents that are of emergent characteristic and impacting life or national security, the provider must report within 72 hours after the discovery of such incident. For other serious incidents, the provider must report within 5 working days. And in either event, a preliminary report must be submitted to the state agency using the forms provided in Decree 142.
 

6.

Domestic Agent Requirement
 
Foreign providers who provide High-Risk AI must have a legal contact point in Vietnam. Further, in cases where such High-Risk AI is subject to mandatory conformity evaluation, the provider must maintain commercial presence or an authorized representative in Vietnam.^[10] This risk-based requirement is different from the approach taken by the Korean AI Act, which mandates domestic representative in Korea only with respect to foreign business operators that meet thresholds of users or revenue as prescribed by the government.^[11]
 

7.

Prohibited Actions & Administrative Penalty

The Vietnamese AI Law specifically prohibits certain practices related to AI, including (i) using AI to commit crimes, (ii) intentionally and systematically deceiving or manipulating human perception and behavior, (iii) exploiting vulnerable groups such as youth and elderly, (iv) creating and disseminating fake content that can cause harm to national or public interest, (v) inhibiting or circumventing safety controls or human oversight mechanisms, and (vi) manipulating, concealing information required to be disclosed for transparency.^[12]
 
Regarding non-compliance, the Vietnamese AI Law promulgates liabilities in forms of administrative fine, civil damage and criminal prosecution; however, no specific amount of administrative penalties was included in the law. This was a significant step-down from earlier drafts of the law, which proposed up to 2 billion VND or 2% total revenue as administrative fine.
 
On the other hand, the Korean AI Act contemplates more specific enforcement mechanism, involving administrative fine of up to 30 million KRW or imprisonment no more than 3 years.
 

8.

Grace Periods

The Vietnamese AI Law came into effect on March 1, 2026 and provides two grace periods for compliance.^[13] For AI system in healthcare, education, and finance, the providers and deployers have 18 months to fulfill their obligations under the law. All other types of AI will have 1 year grace period (i.e., until March 1, 2027). For reference, the current grace period under the Korean AI Act is also at least 1 year.
 

9.

National AI Ethics Framework
 
The National AI Ethics Framework covers ethical standards and values for AI activities in Vietnam, which will be mandatory for AI systems used in the State management and public service. It is also highly encouraged for all other organizations and individuals to apply these ethical standards.
 
Generally, the aims of Vietnamese National AI Ethics Framework are in line with the basic principles in the Korean National Guidelines for AI Ethics, adopted in December 2020 by the Ministry of Science and ICT. The four core Vietnamese ethical principles with respect to AI are as follows.^[14]
 

The Vietnamese National AI Ethics Framework will be reviewed and updated every three (3) years to keep pace with technological and legal changes. The Ministry of Science and Technology provides digital platforms and automated tools for entities to self-evaluate their ethical compliance.
 

10.

Outlook
 
Being the first in South East Asia, even though the Vietnamese AI Law by itself does not include detailed regulations that would be immediately implemented, it still serves as an important framework for Vietnamese government to formulate both incentive policies and specific AI regulations such as the National AI Ethics Framework or Decree 142.
 
Vietnamese law designates national AI infrastructure as a strategic asset; thus, investments in this sector expect to benefit from the highest preferential tax rates and investment policies.^[15] Beyond fiscal support, key incentives^[16] for participants in AI sectors include dedicated mechanism for startups and SMEs, access to the national AI database and regulatory sandboxes, advantages in market procurement, and priority status within the national program. Most importantly, the Vietnamese AI Law and its related regulations sends a strong signal that Vietnam commits to the global standards regarding AI by aligning with the EU AI Act and consequently the Korean AI Act.
 
Not only AI developers or providers, but also the deployers of High-Risk AI in Vietnam (e.g., a commercial bank in Vietnam using AI for credit screening) must be ready to comply with the regulatory requirements including those already promulgated in Decree 142. Other deployers of Medium-Risk or Low-Risk AI (which may include most enterprises incorporating AI in their operations) should be aware of their transparency, labelling and report obligations. In summary, since AI is a constantly developing field, companies conducting business in Vietnam should keep monitoring for important legislation development and accordingly adjust their global compliance policies to the specific regulations of Vietnamese market.