Vietnam’s Artificial Intelligence (AI) Law establishes a comprehensive legal framework governing the development, supply, deployment, and use of AI systems in Vietnam. Built on a risk-based approach, the Law assigns obligations, where applicable, to developers, providers, deployers, and users, proportionate to the risk level posed by an AI system.
This article summarizes the key obligations of each relevant party, with specific references to the corresponding articles and clauses of the AI Law.
1. Risk-Based Classification of AI Systems
Before an AI system is placed into use, the provider or deployer must classify the system by risk level as low, medium, or high, pursuant to Article 9 Article 10.1. Where the risk level cannot be clearly determined, the responsible party must request guidance from the Ministry of Science and Technology (MOST) in accordance with Article 10.4.
For medium- and high-risk systems, the classification results must be notified to MOST via the national one-stop electronic portal (Article 10.3). If an AI system is modified, upgraded, or integrated in a manner that creates new risks or increases the risk level, reclassification is required under Article 10.2.
2. General Obligations Applicable to All AI Systems
🔑 Safety, Security, and Reliability
All AI systems, regardless of risk level, must ensure safety, security, and reliability throughout their lifecycle, as mandated by Article 12.1. Developers, providers, and deployers share responsibility for implementing appropriate technical and organizational measures.
In the event of a serious incident, responsible parties must apply technical measures, including temporary suspension or withdrawal of the AI system, and notify competent authorities in accordance with Article 12.2(a). Incidents must also be recorded and reported in a timely manner pursuant to Article 12.2(b).
🔑 Transparency and Accountability
Transparency must be maintained throughout the development, provision, and operation of AI systems (Article 11 and Article 13.5). Relevant parties are required to:
- Provide explanations and information to state authorities upon request (Article 14.1(e));
- Disclose risk warnings and necessary information to users and affected individuals (Article 14.1(e));
- Operate AI systems strictly for their intended purposes and avoid generating new risks (Article 14.2(a)).
🔑 Lawful Use of AI
All parties, including end users, are prohibited from using AI systems to conduct unlawful activities, in accordance with Article 7 and Article 29 of the Law.
3. Obligations for High-Risk AI Systems
High-risk AI systems are subject to enhanced regulatory controls reflecting their potential impact on fundamental rights, public safety, and social order.
🔑 Conformity Assessment and Certification
Prior to deployment, high-risk AI systems must undergo a conformity assessment pursuant to Article 13, and:
- The system shall obtain a conformity certification as required by law; or
- Where the system is not subject to mandatory third‑party certification, the provider may perform an internal conformity assessment or may engage a competent, accredited conformity assessment body (or other body designated/recognised under applicable law).
Foreign providers must maintain a commercial presence, authorized representative, or contact point in Vietnam, as required therein.
🔑 Ongoing Compliance and Risk Management
Providers of high-risk AI systems must maintain conformity and publicly disclose relevant information throughout operation (Article 13.3). They are also required to establish and maintain an AI risk management system, ensure proper governance of training, testing, and operational data, and retain technical documentation and operational logs as outlined in Article 14.
🔑 Human Oversight and Technical Standards
High-risk AI systems must ensure effective human oversight, including the ability for human intervention and control, in accordance with Article 14.2(b) and Article 14.1(d). During operation, such systems must continuously comply with applicable AI standards and technical regulations (Article 14.2(c)).
By combining a risk-based regulatory structure with defined obligations, Vietnam’s AI Law creates a coherent framework for responsible AI governance. Organizations involved in AI activities should prioritize early compliance, particularly in relation to risk classification, conformity assessment, documentation, and ongoing risk management, to ensure compliance.
Connect with Humane-AI Asia to anticipate regulatory change, operationalize AI governance, and leverage compliance as a foundation for responsible and sustainable innova
For more information about our services, visit: https://humane-ai.asia/en
Humane-AI Asia
Tran Vu Ha Minh | 0938801587
