As a general-purpose technology, artificial intelligence not only enhances humanity's physical capacity to transform the world but also expands the intellectual boundaries of human cognition, bringing unprecedented opportunities for the development of education. However, the deep integration of AI into the education sector, while bringing transformative opportunities for teaching, research paradigms, and governance models, also harbors ethical risks such as the degradation of cognitive agency, educational imbalance, and privacy breaches.

In 2026, under the guidance of the Department of Science, Technology, and Information Technology of the Ministry of Education, scholars from Zhejiang University, Beijing Normal University, and The Chinese University of Hong Kong authored "AI Education Ethics: A Reference Framework." International and domestic experts conducted reviews of the framework through the World Digital Education Alliance Expert Advisory Committee and the Ethics and Governance Committee of the AI Open Alliance. On May 12, at the 2026 World Digital Education Conference in Hangzhou, Zhejiang, "AI Education Ethics: A Reference Framework" was officially released as a major outcome.

"AI Education Ethics: A Reference Framework" establishes the core concepts of "human-centered agency, collaborative symbiosis, context-appropriate beneficence, and classified governance." It takes the holistic development of humans as the fundamental goal, the, interaction among "teachers, students, and machines" as the core mechanism, the appropriateness to educational contexts as the value measure, and the classified application and governance across multiple educational stages and scenarios as the practical path, systematically constructing an ethical system for the AI era in education. It presents four dialectically unified basic behavioral guidelines: "strengthen human-machine collaboration boundaries to highlight the value of educational agency; accurately adapt to educational scenarios and improve classified governance mechanisms; build a solid data security defense line and strictly uphold the bottom line of privacy protection; promote algorithmic fairness and transparency, and establish sound accountability mechanisms." This reference framework clarifies three types of risks in AI education applications along with principles of responsibility attribution: misuse risks and subjective fault attribution, failure risks and objective defect attribution, and systemic risks and ecological governance attribution. Additionally, the framework covers ethical behavioral norms for educators, learners, and educational institutions across basic education, higher education, and vocational education, clearly defining three types of behavioral boundaries and their dynamic adjustment mechanisms: drawing "prohibited access" red lines to safeguard educational baselines, regulating "limited use" boundaries to guide human-machine collaboration, and exploring "encouraged use" spaces to stimulate innovative potential. The framework emphasizes that educational equity is a core issue in AI education, and that the promotion of AI education applications should always prioritize educational equity.

Click here for more details.