Vietnam's Law on Artificial Intelligence has been in force since 1 March 2026. Since then the government has added a guiding decree, an official list of high-risk systems and an online registration portal.

Of the ASEAN economies compared here, Vietnam is the only one with a standalone AI statute in force. Whether that is a "head start" depends on enforcement, and key pieces are still missing.

What the National Assembly passed

The National Assembly approved Law No. 134/2025/QH15 on Artificial Intelligence on 10 December 2025, with 429 of 434 deputies present voting in favour, according to VnEconomy. The law has 35 articles and took effect on 1 March 2026, according to the government's news portal. Its core principle is that AI must serve humans, not replace them.

The Ministry of Science and Technology (MOST) leads. The law replaces the AI provisions of the Law on Digital Technology Industry, Baker McKenzie notes.

The main provisions

Three risk tiers. According to Duane Morris, high-risk systems need conformity assessments before they go to market, ongoing risk management, human oversight and registration on the national portal. Medium-risk systems face monitoring, sample audits and transparency duties. Low-risk systems face only light oversight.

Foreign providers. Foreign providers of high-risk systems needing pre-market assessment must set up a commercial presence or appoint an authorised representative in Vietnam, Duane Morris notes.

Labelling. People must be told when they are dealing with an AI system. AI-generated audio, images and video must carry machine-readable marks. Content that imitates real people, voices or events needs clear labels, according to Baker McKenzie. Banned practices include deceptive simulation of real people or events and targeting vulnerable groups.

Sandbox and incentives. The law sets up a National AI Development Fund (the government later decided not to create it as a separate fund), regulatory sandboxes with simpler procedures, and support for AI clusters in high-tech parks, Baker McKenzie says. The government portal also cites a national AI computing centre and AI vouchers for businesses.

Penalties. Article 29 provides for administrative sanctions, possible criminal liability and civil damages, but sets no fine caps or revenue-based percentages, leaving amounts to decree, according to Baker McKenzie. A June 2026 note from Graf von Westphalen likewise cites no figures. Some online summaries quote specific caps; NDNews could not confirm them in the law or any published sanctions decree.

Transition. Systems already in use before 1 March 2026 have until 1 September 2027 in healthcare, education and finance, and until 1 March 2027 in other sectors.

Implementing rules issued so far

Instrument Date What it does
Decree 142/2026/ND-CP Issued 30 April, effective 1 May 2026 Classification, notification, labelling, incident reporting, transition
Decision 33/2026/QD-TTg Issued 30 June, effective 15 August 2026 Official high-risk list
AI one-stop portal (ai.dda.gov.vn) Launched 28 August 2026 Registration, risk notices, incident reports, pilot trials
National AI Strategy to 2030 Approved 28 August 2026 National targets

Under Decree 142, providers classify systems before deployment and notify MOST of medium- and high-risk ones; serious incidents must be reported within 72 hours if urgent, otherwise five working days, Tilleke & Gibbins says. Operators of existing systems had to file a transition notice by 30 June 2026, according to Viet An Law.

Decision 33 covers six sectors; Allen & Gledhill counts 46 systems, 31 of them in transport and two in banking. Allen & Gledhill gives the issue date as 15 August, while LuatVietnam and Tilleke & Gibbins say 30 June; all agree it took effect on 15 August.

The portal went live on 28 August, Bao Phap Luat reported. The strategy approved the same day targets an AI economy worth about 6% of GDP and US$5 billion in AI-related exports by 2030, according to Viet Nam News. A draft decree on the National AI Development Fund went to consultation until 12 May 2026, but on 17 June Deputy Prime Minister Ho Quoc Dung agreed with MOST's proposal not to create a new fund and to fold AI tasks into existing funds instead, VietTimes reported.

How Vietnam compares

Jurisdiction Approach (as of 7 October 2026)
Vietnam Binding AI law in force; high-risk list and portal in operation
Singapore Voluntary frameworks and the AI Verify testing toolkit
Indonesia Two presidential regulations (roadmap and ethics) awaiting signature
Thailand Draft AI Act; public consultation closed 14 August 2026
Malaysia 2024 ethics guidelines; AI Governance Bill drafted; the minister hopes to table it by Q1 2027
ASEAN Voluntary guide (2024) plus a generative AI annex (2025)
EU AI Act in force; high-risk deadlines pushed back to 2027–2028

Singapore relies on guidance: the AI Verify testing framework (May 2022) and, from 22 January 2026, a voluntary Model AI Governance Framework for Agentic AI, according to Bird & Bird.

Indonesia's two draft presidential regulations, one on an AI roadmap and one on AI ethics, were still awaiting the President's signature in early September. Edwin Hidayat Abdullah, Komdigi's director general for the digital ecosystem, said on 3 September: "All that remains is the President (signing the regulations)" (NDNews translation).

Thailand's Electronic Transactions Development Agency put out a risk-based draft AI Act for consultation in July 2026. Rajah & Tann estimates enactment could take six to 24 months; as of 29 September no Cabinet approval had been reported, according to Thai Examiner.

Malaysia's draft AI Governance Bill is complete. Digital Minister Gobind Singh Deo told Malay Mail on 10 September: "We are hoping to table it this quarter or in the first quarter of next year." On 5 October he said controls would be scaled to "the potential harm of the AI system", The Star reported.

ASEAN launched its voluntary Guide on AI Governance and Ethics on 2 February 2024 and a generative AI supplement on 17 January 2025, according to IMDA.

The EU's Digital Omnibus on AI, in force since 27 July 2026, moved stand-alone high-risk deadlines to 2 December 2027 and embedded-product deadlines to 2 August 2028, according to Lewis Silkin. Vietnam borrows the EU's risk logic, but its general transition deadline for existing systems (1 March 2027) comes earlier, and newly deployed listed systems are covered already.

What it means for businesses and investors

What to watch

  • A sanctions decree setting fine levels
  • How AI support is channelled through existing funds after the decision not to create a separate AI fund
  • Guidance on conformity-assessment methods
  • 1 March 2027 and 1 September 2027 transition deadlines
  • Indonesia's presidential regulations, Thailand's Cabinet review and Malaysia's bill

Key numbers

  • 35 articles in Law No. 134/2025/QH15
  • 429 of 434 deputies present voted in favour on 10 December 2025
  • 1 March 2026: date the law took effect
  • 46 high-risk systems across 6 sectors, as counted by Allen & Gledhill
  • 72 hours: deadline for reporting urgent serious incidents
  • 6% of GDP: Vietnam's 2030 target for the AI economy

This article was researched with AI assistance and reviewed by the NDNews editorial team. It is based on official data and media reports as of 7 October 2026.