42% of Vietnam’s population is already using AI tools. But how many of them know that since March 1, 2026, a new law directly governs how they use those tools - especially in marketing?
Vietnam’s Law on Artificial Intelligence has been in effect for nearly two months. Most analysis has focused on healthcare, fintech, and high-risk AI systems. For marketers using AI daily, here’s what actually matters.
Vietnam’s Three-Tier AI Risk System - Where Does Marketing Fit?
The law establishes three tiers: high-risk, medium-risk, and low-risk.
High-risk covers AI deployed in finance, healthcare, education, and critical infrastructure. These require pre-market government approval and registration in the National AI Database before deployment.
Medium-risk - the tier most relevant to marketers - applies to any AI system that “directly interacts with users or generates content.” This includes: AI content generation tools, AI image and video creation platforms, AI-powered customer service chatbots, and AI personalization engines.
Medium-risk systems must ensure users know they are interacting with AI, and must clearly label outputs.
Low-risk covers AI with minimal societal impact - allowed to self-regulate under general principles.
3 Specific Requirements Content Teams Must Implement
The law does not ban AI. It requires transparency about using AI. Specifically:
1. Label AI-generated content Images, video, and audio that are AI-generated or significantly AI-edited must carry machine-readable labels per government specifications. Content that “could confuse users about authenticity” must have clearly visible notices.
Practical applications: ad banners created with Midjourney, TikTok videos with AI voice-overs, product images enhanced by AI - all fall within scope.
2. Disclose AI interactions to users Chatbots, virtual assistants, or any automated system communicating with customers must notify users that they are interacting with AI - not a human.
3. Human oversight mechanisms For medium and high-risk systems, documented processes for human intervention and AI decision override must exist. Not just “someone reviews outputs” - auditable trails and structured override workflows are required.
The Deadlines That Should Be on Your Roadmap Now
The law took effect March 1, 2026. Two transition windows apply:
- Healthcare, education, finance: 18 months - deadline September 1, 2027
- All other industries - including marketing, advertising, and media: 12 months - deadline March 1, 2027
That’s roughly 11 months from today. Long enough to feel comfortable procrastinating. Short enough to create a compliance crunch if you wait.
The practical compliance list, as documented by Asia Tech Lens (2026): AI inventory with named owners and risk classifications, risk classification memoranda, retrievable audit logs (inputs, outputs, model versions, business actions), vendor change control with notification trails, and documented human oversight mechanisms.
For a mid-size marketing agency running 5-10 AI tools across teams - this is not a one-week project.
The Core Tension: Rushed or Right on Time?
The Business Software Alliance - representing Microsoft, Salesforce, Zoom - warned that implementation was “insufficient for stakeholders to analyze the document rigorously or provide substantive feedback” (TechPolicy Press, 2026). The Computer and Communications Industry Association flagged risks of “compliance bottlenecks” and potential deterrence of foreign market entry.
But Vietnam’s adoption curve tells a different story: 48x increase in AI users from 2023 to 2024 - one of the fastest single-year adoption surges globally. 65% of SMEs already using AI tools (InvestVietnam, 2026).
This is the classic tension of an emerging market where adoption outpaces governance. The AI Law was designed to close that gap. The real question is whether the implementing decrees - still being drafted - will be specific enough for businesses to act on.
Maximum penalties: VND 2 billion (~US$75,800) for organizations, VND 1 billion for individuals (IAPP, 2026). Significant enough to take seriously, particularly for small and mid-size agencies.
Who on Your Marketing Team Needs to Act Now?
Content creators using Midjourney, Stable Diffusion, or Adobe Firefly for ad assets - need to establish an AI labeling workflow for all output.
Social media managers using AI for captions or AI voice-over for Reels and TikTok - need a clear AI disclosure policy.
Email marketers using AI personalization engines, especially where emails “simulate” a specific person’s voice - fall within scope.
Agency owners deploying chatbots or virtual assistants for clients - must ensure end-users are notified they are interacting with AI.
Not every tool requires national-level registration. But the transparency and labeling requirements apply broadly. And the 11-month window to build compliant workflows is shorter than it looks.
NateCue's Take
Most analysis of Vietnam's AI Law focuses on healthcare and fintech. But marketing is the sector adopting AI fastest - 93% of marketers use AI for content, 42.5% for content creation specifically (Digital Applied, 2026). This is the first time content marketing in Vietnam has real compliance risk. Not "best practice." Not "platform policy." Law, with fines up to VND 2 billion. Marketing teams need to sit down with legal and audit their AI tool inventory now - not in February 2027. The bigger picture: this law creates a level playing field. Agencies that proactively disclose AI-generated content will build trust faster than those that don't. Transparency isn't just compliance burden - it's a brand differentiator waiting to be claimed.