AI Regulation Tracker / Regulation signed
Vietnam Names Its High-Risk AI Systems in Decision 33/2026 (Quyet dinh 33/2026/QD-TTg)
Regulatory summary: Vietnam issued Decision 33/2026/QD-TTg on July 2, 2026, the first list naming which AI systems are legally high-risk across education, ethnic/religious affairs, healthcare, banking, litigation, and transport under AI Law 134/2025. It takes force August 15, 2026.
The Prime Minister's office has, for the first time, enumerated which AI systems are legally high-risk across six regulated sectors.
Key takeaways
- Vietnam moved from an abstract statutory risk tier to a concrete, named list. Operators no longer self-assess in a vacuum; a specific function now triggers high-risk status by law.
- Banks and lenders (automated credit decisioning, automated electronic transactions), hospitals and health-tech, courts and legal-tech serving litigation, edtech and schools, transport/autonomous-driving operators, and their in-house and outside counsel.
- Status: Signed.
- Map each production AI system against the six-sector list; if listed, calendar the applicable compliance deadline (Mar 1 or Sep 1, 2027) and begin the high-risk conformity build now.
Vietnam has, for the first time, told operators exactly which artificial-intelligence systems the state considers legally high-risk. On July 2, 2026, Deputy Prime Minister Ho Quoc Dung signed Quyet dinh 33/2026/QD-TTg (Decision 33/2026/QD-TTg), issuing the Danh muc he thong tri tue nhan tao co rui ro cao, the List of High-Risk AI Systems, under Luat 134/2025 (AI Law No. 134/2025). The list takes force on August 15, 2026. For any organization running AI in Vietnam or selling AI into it, this is the document that converts an abstract statutory category into a concrete, function-by-function checklist.
| Date | Jurisdiction | Rule | Affected professionals | Status or effective date |
|---|---|---|---|---|
| 2026-07-09 | Vietnam | Vietnam moved from an abstract statutory risk tier to a concrete, named list. Operators no longer self-assess in a vacuum; a specific function now triggers high-risk status by law. | Banks and lenders (automated credit decisioning, automated electronic transactions), hospitals and health-tech, courts and legal-tech serving litigation, edtech and schools, transport/autonomous-driving operators, and their in-house and outside counsel. | Signed; in force Aug 15, 2026 |
Frequently Asked Questions
Does Decision 33/2026/QD-TTg apply to my company if we are not headquartered in Vietnam?
The list turns on the AI function and the Vietnamese market, not on where you are incorporated. If you provide or deploy a listed high-risk AI system in Vietnam, the high-risk duties under AI Law 134/2025 and Decree 142/2026/ND-CP apply. Foreign providers selling automated credit, health, education, transport, or justice tools into Vietnam should assume coverage and verify against the operative list.
My bank's AI only scores applicants and a human makes the final loan decision. Are we still high-risk?
Very possibly. The list expressly reaches systems that automatically determine credit approval and systems that automatically execute electronic transactions. A human sign-off does not automatically remove a system from scope; the regime's emphasis on human oversight and intervention suggests such controls are part of compliance, not an exemption from it. Review your specific system against the Decision's text and the AI Law.
How many AI systems are on the list?
The Decision organizes high-risk systems across six sectors. Transport is the largest single category with 31 named systems, which we confirmed against the primary source. Vietnamese reporting describes a total in the mid-40s (near 46). Because the exact overall count was not stated verbatim in the Government Portal excerpt we verified, treat the operative Decision text as the authoritative count.
When must existing AI systems comply?
The Decision is in force August 15, 2026. Systems already operating before that date must comply before March 1, 2027 for most sectors, including transport and litigation, and before September 1, 2027 for healthcare, education, and finance. New systems entering the market are governed from the in-force date.
What is the relationship between Decision 33 and AI Law 134/2025?
Law 134/2025 is the statute; it took effect March 1, 2026 and created the high-risk category in the abstract. Decision 33/2026/QD-TTg is the implementing instrument that names which concrete systems fall into that category for six sectors. The law and Decree 142/2026/ND-CP supply the obligations; the Decision tells you when a given system triggers them.
Internal links
Sponsored Training
Browse the full AI Regulation News tracker
Informational analysis for working professionals, not legal advice. Confirm how any rule applies to your situation with qualified counsel.