AI Regulation Tracker / New law

Vietnam's AI Law Decree 142 Is Live: Risk Tiers, Content Labelling, 72-Hour Incident Reports and a National AI Portal

Regulatory summary: Vietnam's Decree 142/2026/ND-CP, issued by the Government on April 30, 2026 and effective May 1, 2026, is the principal decree detailing and implementing the Law on Artificial Intelligence (Law No. 134/2025/QH15, effective March 1, 2026).

Decree 142/2026/ND-CP is the master implementing rule for Vietnam's Law on Artificial Intelligence. It sets a three-tier risk system, mandatory labelling of AI-generated media, serious-incident reporting within 72 hours, and a national AI portal and systems database run by the Ministry of Science and Technology. Systems already operating before March 1, 2026 had to file a transition notice within 60 days.

Primary source

By Anthony Guerriero, Founder, The Leveraged Years · Reviewed by The Leveraged Years Editorial Desk · Published July 9, 2026 · Last updated July 9, 2026

Vietnam's AI Law Decree 142 Is Live: Risk Tiers, Content Labelling, 72-Hour Incident Reports and a National AI Portal regulation briefing

The Leveraged Years AI Regulation Tracker

Key takeaways

Vietnam moved from a framework AI Law to operative rules. The decree fills in how the law works in practice: it defines the three risk levels and puts the classification duty on the provider before a system is deployed, sets concrete labelling obligations for synthetic media, fixes incident-reporting timelines, and builds the national portal and database that make classification and identification codes real. It also sets transitional duties for systems already in operation when the law took effect.
AI developers and vendors selling into Vietnam; enterprises deploying AI in customer-facing or high-impact processes; compliance, legal and data-protection officers at multinationals operating in Vietnam; and platforms distributing AI-generated audio, image or video to Vietnamese users.
Status: In force.
Build an inventory of every AI system touching Vietnamese users, classify each as high, medium or low risk, confirm labelling is applied to synthetic audio, image and video that could mislead, wire up a 72-hour serious-incident reporting path, and verify that any system operating before March 1, 2026 filed its transition notice on the national portal before June 30, 2026.

Regulatory Briefing

Bindingness: Binding government decree in force. Decree 142/2026/ND-CP is a Nghi dinh, a legally binding implementing decree issued by the Government under the Law on Artificial Intelligence. It is not draft guidance. It took effect on May 1, 2026 and carries detailed duties for AI providers and deployers.
Scope: Providers and deployers of AI systems in Vietnam, across the risk tiers defined by the AI Law and the decree. Certain systems are carved out of public-disclosure duties, including AI serving internal Party and political operations, classified national defence and security systems, and national-security databases. Purely internal-use, technical-editing, text-processing and research or testing uses are exempt from some labelling duties.
What Changed: Vietnam moved from a framework AI Law to operative rules. The decree fills in how the law works in practice: it defines the three risk levels and puts the classification duty on the provider before a system is deployed, sets concrete labelling obligations for synthetic media, fixes incident-reporting timelines, and builds the national portal and database that make classification and identification codes real. It also sets transitional duties for systems already in operation when the law took effect.
Who This Affects: AI developers and vendors selling into Vietnam; enterprises deploying AI in customer-facing or high-impact processes; compliance, legal and data-protection officers at multinationals operating in Vietnam; and platforms distributing AI-generated audio, image or video to Vietnamese users.
Current Status: In force. Issued April 30, 2026, effective May 1, 2026. The transition-notice window for pre-existing systems ran 60 days from the effective date, closing before June 30, 2026, with the national AI portal becoming the channel for classification notifications and identification codes.
Effective Date: May 1, 2026. Transition notice for systems operating before March 1, 2026 due within 60 days of the effective date (before June 30, 2026).
Primary Source: https://vanban.chinhphu.vn/?docid=218029&pageid=27160
Secondary Analysis: Tilleke and Gibbins, Atsumi and Sakai, and Viet An Law published detailed English-language analyses in May 2026; MLex reported on the finalised decree and the incident-reporting timelines. Vietnamese state media (Nhan Dan, chinhphu.vn media portal) reported the three-tier classification and the national portal.
Workflow Impact: AI providers must classify systems by risk before launch and stand behind the accuracy of that classification, apply machine-readable provenance markings and user-facing labels to AI-generated media that could mislead, register or self-declare systems through the national portal to obtain identification codes, and operate an incident-reporting process that can produce a preliminary report within 72 hours for serious incidents.
Operational Next Step: Build an inventory of every AI system touching Vietnamese users, classify each as high, medium or low risk, confirm labelling is applied to synthetic audio, image and video that could mislead, wire up a 72-hour serious-incident reporting path, and verify that any system operating before March 1, 2026 filed its transition notice on the national portal before June 30, 2026.
Last Verified: July 10, 2026

Date	Jurisdiction	Rule	Affected professionals	Status or effective date
2026-07-09	Vietnam	Vietnam moved from a framework AI Law to operative rules. The decree fills in how the law works in practice: it defines the three risk levels and puts the classification duty on the provider before a system is deployed, sets concrete labelling obligations for synthetic media, fixes incident-reporting timelines, and builds the national portal and database that make classification and identification codes real. It also sets transitional duties for systems already in operation when the law took effect.	AI developers and vendors selling into Vietnam; enterprises deploying AI in customer-facing or high-impact processes; compliance, legal and data-protection officers at multinationals operating in Vietnam; and platforms distributing AI-generated audio, image or video to Vietnamese users.	In force. Issued April 30, 2026, effective May 1, 2026. The transition-notice window for pre-existing systems ran 60 days from the effective date, closing before June 30, 2026, with the national AI portal becoming the channel for classification notifications and identification codes.

Source File

Instrument: Decree No. 142/2026/ND-CP detailing and implementing the Law on Artificial Intelligence
Authority: Government of Vietnam (implementing Law No. 134/2025/QH15; portal and database administered by the Ministry of Science and Technology)
Date: Issued April 30, 2026; effective May 1, 2026; transition notice due before June 30, 2026
Citation: Nghi dinh 142/2026/ND-CP, Chinh phu, 30 April 2026
Primary Source: https://vanban.chinhphu.vn/?docid=218029&pageid=27160
Corroboration: Tilleke and Gibbins, Vietnam Takes Major Step in AI Governance with New Guiding Decree; Atsumi and Sakai, Key Highlights of the Decree Implementing the Law on Artificial Intelligence (28 May 2026); Viet An Law, Decree 142/2026/ND-CP guiding the Artificial Intelligence Law in Vietnam; MLex, Vietnam finalizes AI implementation decree, relaxes incident-reporting timelines; Nhan Dan and chinhphu.vn media portal (state media) on the three-tier classification and national AI portal.

Frequently Asked Questions

Is Decree 142/2026/ND-CP in force?

Yes. The Government issued it on April 30, 2026 and it took effect on May 1, 2026. It is a binding implementing decree for the Law on Artificial Intelligence, published on the official government legal database.

How many risk tiers are there, and who decides?

Three: high, medium and low. The AI provider must classify a system before it is deployed and is legally responsible for the accuracy of that classification, reassessing if the deployment or risk changes.

When must AI-generated content be labelled?

Providers must apply machine-readable markings to AI-generated or AI-modified audio, image and video, and deployers must give clear notifications where the content could mislead users about authenticity, especially where it mimics a real person or recreates factual events. Internal-use, technical-edit, text-processing and research uses are exempt from the display-labelling duty.

What is the incident-reporting deadline?

A preliminary report within 72 hours of discovering an urgent serious incident, and within five working days for other serious incidents, with a follow-up report to follow.

What was the transition deadline for existing AI systems?

Systems operating before March 1, 2026 had to file a transition notice within 60 days of the decree's effective date, meaning before June 30, 2026, through the national AI portal.