Decision No. 33/2026/QD-TTg is a decision of the Prime Minister of Vietnam, signed by Deputy Prime Minister Ho Quoc Dung and issued June 30, 2026, promulgating Vietnam's List of High-Risk Artificial Intelligence Systems together with operating principles and a compliance roadmap for the listed systems. It takes effect August 15, 2026 (Source: vietnam.vn).
Scope and classification
The decision classifies 46 AI systems as high-risk, spanning education, healthcare, banking, litigation, transportation, and ethnicity-and-religion applications (Source: mst.gov.vn). Vietnamese state-media coverage attributes the largest share of listed systems — 31 of the 46 — to the transportation sector (Source: vietnam.vn). Reported counts for the remaining sectors vary across secondary accounts, so a per-sector breakdown is not reproduced here.
The list-based approach designates specific systems for stricter management rather than defining high-risk categories abstractly, a structural difference from the EU AI Act's Annex III category model. State-media reports describe the decision as implementing Vietnam's Law on Artificial Intelligence, alongside Decree No. 142/2026/ND-CP, and as part of completing the country's legal framework for managing AI "in a safe, transparent, and responsible manner" (Source: vietnam.vn). The Ministry of Science and Technology is the responsible ministry (Source: vietnam.vn).
Compliance roadmap
Although the decision enters into force August 15, 2026, listed systems are subject to staged compliance deadlines of March 1, 2027 and September 1, 2027 depending on sector (Source: mst.gov.vn). State-media accounts describe the accompanying operating principles as requiring providers of listed systems to comply with legally stipulated operating requirements, including human oversight and intervention (Source: vietnam.vn).
Regional context
The decision places Vietnam among the Asia-Pacific jurisdictions adopting risk-tiered AI regulation in 2025–2026, alongside the South Korea AI Basic Act and Japan's AI Promotion Act, and contemporaneous with China's application-specific instruments such as the Anthropomorphic Interaction Measures. It was issued in the same week as several other Asia-Pacific regulatory moves, including South Korea's Personal Information Protection Commission 3rd Basic Plan (2027–2029) and Hong Kong's personal-data AI sandbox for schools (Source: techieray.substack.com).
Relationships
- instance-of: Risk-Based AI Regulation — a list-based national implementation of risk-tiered AI regulation
- related: EU AI Act (Regulation 2024/1689) — contrasting category-based high-risk classification model
- related: South Korea AI Basic Act (Framework Act on AI Development and Trust, 2024/2026), Japan AI Promotion Act (Act on Promotion of Research, Development and Utilisation of AI-Related Technologies, 2025) — contemporaneous Asia-Pacific AI framework laws