Vietnam orders schools to spell out exactly when AI use is allowed

Vietnam's Ministry of Education and Training (Bộ GD&ĐT) published Official Dispatch No. 6020/BGDĐT-KHCNTT on September 8, 2026, requiring every school in the country to draw up its own rules on the use of artificial intelligence. The document, tied to digital transformation tasks for the 2026-2027 school year, explicitly orders schools to sort AI uses into three categories: allowed, restricted, and fully banned.

What the new dispatch requires

Under the circular, schools must clearly spell out the cases in which students, teachers and administrators may use AI tools, when use is allowed only within limits, and when it is forbidden outright. Schools must also define user responsibilities and ensure data security, transparency in how systems operate, and the ability to verify AI-generated content.

The ministry lists specific, permitted uses of AI for teachers: preparing lessons, creating teaching materials, designing classes, and support with testing and grading. School administrators, meanwhile, may use AI for data collection and analysis, forecasting, and administrative work.

A line AI cannot cross

The ministry is firm on one point: AI does not replace the professional judgment and decision-making responsibility of teachers and school administrators. Any output generated by AI must be checked for accuracy and relevance before it reaches students or is used to make a decision.

Before rolling out any AI tool more broadly, a school must evaluate its quality, effectiveness, safety, and impact on teachers' work. The requirement is meant to stop schools from deploying chatbots or content generators at scale without first checking how they hold up on real classroom tasks.

Part of a wider reform

September's dispatch is not an isolated move but another piece of a plan Vietnam has been assembling for months. Earlier, on June 30, 2026, the ministry issued circular 49/2026/TT-BGDĐT, in effect since August 15, which explicitly bans using AI to cheat on tests, plagiarize, falsify research data, or have someone else's work passed off as one's own without disclosing AI involvement. Learning management systems are required to provide user authentication and activity tracking to support this.

In parallel, a nationwide AI education program is launching that will cover all general-education schools in the country starting with the 2026-2027 school year. The program splits into a mandatory track of 12 class hours per year at every grade level, and an extended track tailored to specific students' needs. The curriculum is organized around four competency areas: human-centered thinking, AI ethics, AI techniques and applications, and AI system design.

A Polish perspective

In Poland, the debate over regulating AI in schools is only just getting started. Cyfrowa Polska (a Polish digital-industry association) has proposed national rules on AI use in schools to the Ministry of National Education, but binding, nationwide regulations are still missing, and in practice it is individual schools and teachers who decide which tools to allow. Vietnam's approach, combining a top-down, mandatory AI curriculum with a clear split between allowed, restricted and banned uses, shows one possible model in which the state sets the framework instead of leaving the decision entirely to school principals.

Other countries in the region are grappling with similar dilemmas. Singapore, France and China have adopted different strategies for student access to artificial intelligence, while Norway opted for a total ban on generative AI in primary schools. Vietnam chose a third path: it does not ban AI, but forces every school to write its own rulebook and take responsibility for it.

For education systems worldwide, including Poland's, the Vietnamese case is a test of whether detailed administrative guidelines actually translate into controlled AI adoption in classrooms, or whether they stay on paper while teachers and students keep using chatbots outside official school channels anyway.