Vietnam's 2025 AI Law Puts People First: Transparency, User Alerts, and a National AI Database

Vietnam's Draft AI Law: What Legal Teams Need to Prepare For

Vietnam is advancing a comprehensive AI Law with a clear anchor: transparency and labeling. At the Vietnam AI4VN 2025 conference on September 26, Deputy Minister of Science and Technology Bui Hoang Phuong stated that users must be notified before engaging with AI systems.

The Ministry aims to submit the draft to the National Assembly by the end of 2025. If passed that year, Vietnam would join a short list of countries with a clear, end-to-end legal framework for AI. The goal is regulatory clarity that also accelerates responsible AI adoption.

Five principles guiding the law

• People-first orientation
• Safety and transparency
• National sovereignty with global integration
• Inclusive and sustainable development
• Balanced, harmonious governance

Transparency and labeling: expected obligations

The draft requires clear notification before users interact with AI systems. This responds to a core risk: people struggling to distinguish human content from AI output.

Expect practical duties to include visible notices at the point of interaction, plain-language disclosures, and potentially persistent indicators for AI-generated content. Client-facing products, call centers, chatbots, and internal decision support tools will need review.

High-risk oversight: national AI database

A standout proposal is a national AI database for monitoring, managing, and ensuring transparency of high-risk systems. While the scope will be defined in drafting, organizations operating high-impact models should prepare for registration, reporting, and documentation duties.

Counsel should inventory current AI use, identify applications that affect rights, critical services, or public interests, and begin building evidence trails for risk assessments and model governance.

National AI infrastructure: two pillars

• Support for innovation and research
• Enablement of government administration

This signals investment in shared capabilities alongside regulatory measures. Public-private coordination will likely be central to implementation.

Ethics and accountability across sectors

The Ministry highlighted education, healthcare, finance, and public administration. AI is encouraged as an assistive tool-especially for students-supporting independent reasoning instead of replacing it.

Healthcare priorities include protecting personal health data and ensuring fair access to services. Financial services and public-sector deployments can expect heightened scrutiny on explainability, redress, and data governance.

Global context

Over 200 former heads of state, diplomats, Nobel laureates, and AI experts have called for a "red line" agreement that sets non-negotiable limits for AI by the end of 2026. Vietnam's approach lands in that global push for clear guardrails with practical enforcement.

As Minister Nguyen Manh Hung put it: "We must develop AI that is fast, safe, and humane. AI must serve people, benefit people - not replace them."

Action checklist for legal and compliance teams

• Map AI touchpoints where users interact with automated systems; draft standardized notices and labeling policies.
• Classify use cases that could be high-risk and begin maintaining technical documentation, data lineage, and evaluation records.
• Update procurement and vendor contracts with transparency, data protection, and potential registration/reporting clauses.
• Strengthen governance for sensitive sectors: student data, health data, financial decisioning, and public services.
• Plan training for product, data, and policy teams on disclosure, testing, monitoring, and incident response.
• Prepare for consultations in 2025 and track implementing decrees that may follow the primary law.

Timeline and next steps

Drafting is underway, with submission to the National Assembly targeted by end-2025. Organizations should begin gap assessments now; retrofitting transparency and documentation late in the process is costly and risky.

Expect further detail on definitions, scope of high-risk systems, and supervisory mechanisms during the drafting cycle. Multi-stakeholder coordination-state, enterprises, and civil society-will be key to workable rules.

Enable internal readiness

If you're building an internal upskilling track for counsel, risk, and product teams, you can review curated options by job role here: Complete AI Training: Courses by Job.