Beijing Releases 2026 AI-in-Education Guidelines With Three-Tier List on Where AI Is Banned, Limited or Encouraged
Beijing has released the 2026 edition of its AI Application Guidelines for Education, adding a three-tier classified list that tells schools exactly where AI is prohibited, where it may be used only under conditions, and where it is encouraged.
The Guidelines were published on September 10 at the 2026 China International Fair for Trade in Services (CIFTIS). The new version is built on a "1+5+2" structure: one overarching guideline setting the framework and compliance requirements; five sector-specific implementation guides covering basic education, higher education, vocational education, lifelong education and special education; and two thematic modules on AI literacy education and ethical safety.
It continues to deepen the seven application scenarios Beijing has promoted since 2024 — AI for teaching, learning, assessment, whole-person development, research, management and infrastructure — now expanded into 66 typical scenarios supported by 150 cases that are expected to be applied in real educational tasks.
The headline addition is the classified application list. Under the prohibited-use list, AI may not replace a teacher's classroom leadership, emotional interaction with students, value guidance or immediate professional judgment. Exercises, tests, writing practice and lab reports that teachers require students to complete independently on the spot may not be generated by AI. In assessment, AI-generated content cannot be presented as a student's own work in comprehensive quality evaluation, academic proficiency tests or admissions-related evaluation, and an AI score cannot be the sole basis for a grade or a developmental judgment.
Under the limited-use list, students must first form their own ideas, judgments or outlines, and teachers must specify which tools are allowed, when they may be used and what evidence of thinking must be submitted. AI output can serve only as a reference, to be verified, questioned and reworked by a human before use.
The encouraged-use list backs AI for open inquiry, cross-disciplinary practice, debate preparation, source retrieval, data summarisation and the creation of interactive courseware and virtual experiments, provided the material is reviewed by teachers.
The Guidelines also set age-graded rules for minors. Children aged 0 to 3 are advised not to have direct contact with AI tools, toys or AI-generated content. Children aged 3 to 6 may have limited experience in areas such as music and art with a parent present; other uses are discouraged, and anthropomorphic emotional-companionship apps are prohibited to prevent dependency.
Primary school students are advised not to use AI alone, though they may use it when a teacher is using AI to support instruction — lower grades focusing on perception and experience, upper grades on understanding. Junior high students may explore it with a teacher or parent present, and senior high students may use it, with a teacher's permission, for inquiry-based learning that engages with how the technology works.
Beijing published the first version of the Guidelines in October 2024, widely seen as the country's first systematic full-stage plan for AI in education. The 2025 edition introduced a four-dimension "North Star – Horizon – New Channel – Traffic Light" framework and a "1+4" structure, and added "AI for infrastructure" as a new scenario.
The Guidelines were developed with the Beijing Academy of Educational Sciences, Beijing Normal University, the Beijing Academy of Artificial Intelligence and Beijing Zhongguancun Academy. City officials say AI application coverage among Beijing primary and secondary students has reached 91%, and Beijing has been approved to build a national pilot base for AI application in education.