Other Briefings
46 High-Risk AI Systems to Face Enhanced Regulatory Oversight in Vietnam
Following the introduction of its first Artificial Intelligence (AI) Law, Vietnam has issued its first official list of 46 high-risk AI systems, providing greater clarity on the technologies subject to enhanced regulatory oversight.
Key takeaways
- Vietnam has published its first official catalogue of 46 high-risk AI systems, covering six sectors: transport, ethnicity and religion, education, healthcare, banking, and judicial proceedings.
- Providers of high-risk AI systems must comply with enhanced governance requirements under the AI Law, including conformity assessment, technical documentation, risk management, and human oversight.
- Existing AI systems benefit from transitional compliance periods, with deadlines ranging from March 1, 2027, to September 1, 2027, depending on the sector.
- Businesses developing or deploying AI in Vietnam should assess whether their systems fall within the designated categories and prepare for the applicable compliance obligations.
The list, issued under Decision No. 33/2026/QD-TTg dated June 30, 2026, identifies AI systems that may pose significant risks to life, health, the lawful rights and interests of individuals and organizations, public interests, national interests, or national security.
Providers and operators of these systems will be required to comply with stricter governance and conformity assessment requirements before the systems can be placed into service.
High-risk AI systems span six sectors
The list covers six sectors:
- Transport – 31 systems;
- Ethnicity and religion – 7 systems;
- Education – 3 systems;
- Healthcare – 2 systems;
- Banking – 2 systems; and
- Judicial proceedings – 1 system.
Transport accounts for more than two-thirds of all high-risk AI systems, reflecting the government’s focus on technologies that could directly affect public safety and critical infrastructure.
Key high-risk AI systems by sector
Education
High-risk AI systems in the education sector include:
- AI systems providing self-learning content using uncontrolled data sources;
- AI systems that automatically assess, grade, or rank students; and
- AI systems that monitor or analyze learner behavior using biometric data, such as facial recognition or eye tracking.
Ethnicity and religion
The list covers AI systems used to:
- Score, classify, or rank applications for ethnic policy benefits;
- Verify information relating to ethnicity or religion and automatically determine its validity;
- Approve or reject administrative applications without human review;
- Allocate public resources or policy support using algorithmic decision-making; and
- Classify individuals or communities based on ethnicity, belief, or religion for regulatory purposes.
Healthcare
Two categories of AI systems are designated as high risk:
- AI-assisted surgical systems; and
- AI-integrated surgical robots capable of directly performing medical interventions or automatically controlling treatment equipment.
Banking
In the banking sector, high-risk AI systems include:
- AI systems that automatically execute high-value electronic banking transactions without human approval; and
- AI-powered credit scoring and lending systems that make automated credit decisions.
Judicial proceedings
The judicial sector includes:
- Large-scale biometric identification systems used in resolving public-interest civil cases.
Transport
Transport represents the largest category, covering 31 AI systems across road, rail, and aviation.
Examples include:
- High-level autonomous driving systems;
- AI systems controlling road and railway traffic signals;
- AI systems managing transport infrastructure and other critical infrastructure;
- AI systems automatically detecting infrastructure failures and ordering operational restrictions;
- Automated air traffic management systems;
- AI decision-support systems for air traffic control;
- AI systems predicting aircraft trajectory conflicts; and
- AI systems automatically allocating airport slots and aircraft gates.
Compliance requirements for high-risk AI systems
Under Vietnam’s AI Law, effective from March 1, 2026, AI systems are classified into three risk categories:
- High risk;
- Medium risk; and
- Low risk.
High-risk AI systems are subject to enhanced compliance obligations, including:
- Conducting conformity assessments before deployment or following significant system modifications;
- Maintaining technical documentation;
- Implementing risk management measures throughout the system’s lifecycle;
- Ensuring appropriate human oversight and intervention; and
- Meeting transparency and governance requirements prescribed by the AI Law.
Conformity assessment requirements
Not all high-risk AI systems are subject to the same conformity assessment requirements. Under Decision 33/2026, the promulgated AI systems must follow one of two conformity assessment pathways prescribed by the AI Law:
- Mandatory third-party certification: Certain high-risk AI systems designated by the Government must undergo conformity assessment by a registered or recognized conformity assessment body before they can be put into operation.
- Self-assessment or third-party assessment: All other high-risk AI systems may be assessed either through a self-assessment conducted by the provider or by a registered or recognized conformity assessment body.
Transitional period
AI systems already in operation before the decision took effect will benefit from a transitional compliance period.
The full-compliance deadlines are:
- September 1, 2027, for AI systems in the education, healthcare, and financial sectors; and
- March 1, 2027, for AI systems in all other sectors.
Implications for businesses
The publication of Vietnam’s first high-risk AI catalogue marks an important milestone in implementing the country’s AI regulatory framework. Organizations developing, deploying, or operating AI systems should assess whether their products fall within the designated categories and prepare for the additional compliance requirements.
For businesses operating in sectors such as healthcare, banking, education, transportation, and public services, early compliance planning, including conformity assessments, governance frameworks, technical documentation, and human oversight mechanisms, will be essential to meet the AI Law’s requirements before the applicable transition deadlines.
See also: Formalizing Artificial Intelligence Governance: Vietnam’s First AI Law
Setting up a business in Vietnam requires navigating company registration, local approvals, and work permit processes. We help FDI companies by preparing and submitting documentation, coordinating with authorities, and ensuring compliance, so they can start operations smoothly and focus on growth.
About Us
Vietnam Briefing is one of five regional publications under the Asia Briefing brand. It is supported by Dezan Shira & Associates, a pan-Asia, multi-disciplinary professional services firm that assists foreign investors throughout Asia, including through offices in Hanoi, Ho Chi Minh City, and Da Nang in Vietnam. Dezan Shira & Associates also maintains offices or has alliance partners assisting foreign investors in China, Hong Kong SAR, Indonesia, Singapore, Malaysia, Mongolia, Dubai (UAE), Japan, South Korea, Nepal, The Philippines, Sri Lanka, Thailand, Italy, Germany, Bangladesh, Australia, United States, and United Kingdom and Ireland.
For a complimentary subscription to Vietnam Briefing’s content products, please click here. For support with establishing a business in Vietnam or for assistance in analyzing and entering markets, please contact the firm at vietnam@dezshira.com or visit us at www.dezshira.com
- Previous Article Representative Office Compliance in Vietnam: Operations, Tax, and PIT Risks
- Next Article
