Summary
- Passed 10 Dec 2025, in force 1 Mar 2026. 35 articles. Covers research, development, supply, deployment and use of AI in Vietnam; AI used only for defence, security or cipher purposes is excluded (Arts. 1, 34).
- Three risk tiers (Art. 9): high risk (can cause significant harm to life, health, rights, national or public interests), medium risk (can confuse or manipulate users who cannot tell they face AI or AI content), low risk (everything else).
- Notify before launch: providers of medium and high-risk systems notify the Ministry of Science and Technology through the National Single-Window AI Portal (Art. 10.3). A national AI systems database is created (Art. 8.2).
- High risk: conformity assessment before service or after a big change (self-assessment or an assessment body, depending on the list issued by the Prime Minister); risk management, data governance, technical documentation and logs, human oversight, incident reporting (Arts. 12 to 14).
- Transparency and labelling (Art. 11): tell users they are dealing with AI; providers machine-mark AI audio, images and video; deployers clearly label AI content that could be mistaken for real, especially content imitating real people or events.
- Prohibited acts (Art. 7): including manipulative or deceptive synthetic content causing serious harm, exploiting vulnerable groups, unlawful data use, defeating human oversight, and hiding mandatory labels.
Who it applies to
- Vietnamese and foreign organisations and individuals taking part in AI activities in Vietnam (Art. 2). Roles: developer, provider, deployer, user (Art. 3).
- Foreign providers of high-risk systems offered in Vietnam must have a legal point of contact there; if mandatory certification applies, they also need a commercial presence or authorised representative (Art. 14.6).
Key dates
- Passed by the 15th National Assembly, 10th session (Law No. 134/2025/QH15)
- Law takes effect (Art. 34)
- Decision 367/QD-TTg issues the plan to implement the Law (reported by the Government news portal)
- End of 12-month transition for existing AI systems outside health, education and finance (calculated from Art. 35)Upcoming
- End of 18-month transition for existing systems in health, education and finance (calculated from Art. 35)Upcoming
Full date record
| Date | Event |
|---|---|
| 2025-12-10 | Passed by the National Assembly |
| 2026-03-01 | In force |
| 2026-03-03 | Implementation plan, Decision 367/QD-TTg (Government news report) |
| 2027-03-01 | Transition ends for existing systems in most sectors (12 months, Art. 35.1b; date calculated) |
| 2027-09-01 | Transition ends for existing health, education and finance systems (18 months, Art. 35.1a; date calculated) |
Penalties and enforcement
- Administrative sanctions or criminal prosecution, plus compensation for damage (Art. 29.1). Fine levels: Not stated in the source; Government will set them by decree (Art. 29.5).
- During transition, existing systems keep running unless the authority finds serious risk and orders suspension (Art. 35.2).
What it means for an Indian company
- If you sell AI products (credit scoring, KYC, chatbots) to Vietnamese banks, assume finance use cases may be high risk and plan for the 18-month transition ending around Sep 2027.
- Decide your role per product: provider (your brand) or developer for a client's brand. The provider notifies on the portal.
- Appoint a Vietnam point of contact for any high-risk system; budget for a local entity or representative if certification becomes mandatory.
- Build labelling into generative features now: machine-readable marks and visible labels (SC-AI-023 Label AI-generated and synthetic content, SC-AI-022 Disclose AI interaction to users, SC-AI-036 Deepfake and impersonation safeguards).
- Reuse EU AI Act style evidence: risk classification SC-AI-004 AI risk classification and use-case tiering, documentation SC-AI-014 Model documentation (model cards, system cards), human oversight SC-AI-026 Human oversight and override, incident reporting SC-AI-031 AI incident management and regulator reporting, conformity readiness SC-AI-039 Conformity assessment and certification readiness.
- Watch for the Prime Minister's high-risk list and the implementing decrees.
Obligations
Full obligation breakdown coming
We break the most requested policies down into numbered obligations mapped to controls. Leave your details and we’ll tell you when Vietnam AI Law is done.
Glossary
- Provider: puts an AI system on the market under its own name or brand, whoever built it.
- Deployer: uses an AI system under its control for professional or commercial work.
- National Single-Window AI Portal: the online channel for notifications and incident reports.
- Conformity assessment: checking a high-risk system meets the rules before launch.
- QH15: suffix for laws passed by the 15th National Assembly.
Key takeaways
- Binding, risk-based law in force since 1 Mar 2026, with notification on a national portal for medium and high-risk AI.
- Labelling and AI interaction disclosure apply broadly, not only to high risk.
- Foreign high-risk providers need a Vietnam contact; existing systems get 12 or 18 months to comply.
- Fines and the high-risk list depend on decrees still to watch.
Official sources
- Law No. 134/2025/QH15, Government legal documents portal, accessed 2026-10-19
- Official Gazette entry, Law 134/2025/QH15, accessed 2026-10-19
- Viet Nam's Law on Artificial Intelligence, English translation, Government news portal, accessed 2026-10-19
- Ministry of Science and Technology news on passage, accessed 2026-10-19
This is general information, not legal advice. Last verified 19 Oct 2026 against the official source. Laws change; confirm the current position with qualified counsel before acting.