Global AI Education Weekly | September 22, 2026
This week’s signal is not simply wider access to AI tools. Education systems are moving simultaneously on governance, teacher capability, and boundaries for student use.
## Week’s signal
As of September 22, the most consequential global signal in AI and education is a shift from the question of whether to introduce a tool toward questions of who may use it, under what principles, and with what support. UNESCO held Digital Learning Week 2026 at its Paris headquarters from September 8 to 11. A ministerial statement presented on September 8 called for deliberative governance, public accountability, and respect for the rights of learners and teachers in the use of AI in education. UNESCO says that more than 25 education ministers and designated representatives adopted a joint statement.[1][2] It is an important cross-border policy signal, not a legally binding treaty.
UNESCO is also running a global discussion and consultation on education in the age of AI from September through November, with comments due by October 15.[3] For schools, tutoring providers, and higher-education institutions, that timetable is a useful prompt to organize evidence from practice: which uses support learning, where safeguards are needed, and where students and teachers need clearer rights and responsibilities.
## International developments
China’s Ministry of Education said on September 4 that it would implement an “AI + Teacher Development” initiative and launch comprehensive AI-literacy training for teachers through the Teacher Center of the National Smart Education Platform.[4] The public material does not specify participant numbers or a completion deadline. The careful reading, therefore, is that this is an announced direction for capacity building—not evidence that training of all teachers has already been completed.
For boundaries around student use, a September 9 Education Week report said that New York City’s school system restricted elementary- and middle-school students’ use of generative AI, while the Los Angeles Unified School District announced a restriction covering all students.[5] Separately, a nationally representative EdWeek Research Center survey of 651 teachers, conducted in February and March 2026, found that 42% reported no training on classroom use of generative AI, 22% reported multiple training sessions, and 9% reported ongoing training.[6] These are teacher-survey findings, not district-level rates or estimates for every K–12 teacher.
Several related items are best read conservatively so that announcements and collaboration discussions are not mistaken for completed implementation. On June 3, the University of Leicester announced a collaboration with Microsoft to provide full access to Microsoft 365 Copilot for all students and staff. The university materials cited here do not confirm a specific September activation date or an official headcount for those covered.[7] Taiwan’s Ministry of Education said that Seoul National University of Education President Shin-Ho Jang and a delegation visited Taiwan from September 8 to 11 to discuss digital technology and smart-education teacher preparation. AI-competency certification for student teachers was described as an area with room for cooperation, rather than a jointly completed or formally launched certification framework.[8]
At the National University of Singapore, the Artificial Intelligence Centre for Educational Technologies (AICET) is hosted by the NUS School of Computing. Its January 7 event, “Pedagogy First—Teaching With, Not By, AI,” records participation by more than 240 educators and stakeholders.[9] That is a useful pedagogical reference point, not a new September policy. In the United States, a 2026 Gallup–Lumina Foundation study found that 57% of 3,801 surveyed students pursuing associate or bachelor’s degrees said they used AI in coursework at least weekly. The survey was fielded in October 2025 among U.S. students aged 18–59 using a Dynata nonprobability web sample, so it should not be generalized to all college students or all forms of AI use.[10]
## Three questions for education institutions
**First, can governance be expressed as executable use cases?** A simple “allowed” or “banned” rule is rarely enough. Institutions can distinguish among brainstorming, practice, information organization, assessment, academic integrity, and sensitive data. For each context, they can state what students, teachers, and administrators may do, must disclose, or must not do. This approach is more realistic than treating a restriction as a total ban, while keeping rights, accountability, and teaching needs in view.
**Second, is teacher development being treated as sustained support rather than one-off tool training?** China’s announced direction and the EdWeek findings both underline that teacher capability is central to implementation. Training may include prompting, but it should also address learning goals, assessment design, bias and privacy recognition, interpretation of student work, and decisions about when not to use AI. Feedback and revision loops matter as well; completion of one workshop is not the same as durable instructional capability.
**Third, does the adoption narrative separate evidence from aspiration?** External communication should distinguish among an announcement, a pilot, access being made available, and actual activation or training completion. Likewise, room for cooperation after an international visit is not the same as a shared standard in operation, and attendance at one event is not proof of system-wide impact. Periodic review using usage evidence, teacher feedback, learning quality, and equity indicators can keep the conversation anchored in verifiable educational value.
## Conclusion
The common thread this week is that maturity in AI education will not be measured only by tool availability. It will also depend on whether institutions can define rights, responsibilities, capability support, and review mechanisms clearly. For cramschool.ai readers, the practical next step is not to chase a single platform or claim compliance with procurement requirements, certifications, or international standards. It is to build explainable, adaptable local governance and professional-learning processes that serve teaching and learning goals.
## Sources
[1] [UNESCO Digital Learning Week 2026](https://www.unesco.org/en/weeks/digital-learning)
[2] [UNESCO: Education ministers call for education to remain a common good in the age of AI](https://www.unesco.org/en/articles/education-ministers-call-education-remain-common-good-age-ai-unescos-digital-learning-week)
[3] [UNESCO global consultation on education in the age of AI](https://www.unesco.org/en/digital-education/artificial-intelligence/consultation)
[4] [China Ministry of Education: AI + Teacher Development](http://www.moe.gov.cn/fbh/live/2026/78355/mtbd/202609/t20260907_1449538.html)
[5] [Education Week: Can school districts really ban AI?](https://www.edweek.org/technology/ai-is-working-its-way-into-everything-can-school-districts-really-ban-it/2026/09)
[6] [EdWeek Research Center: More schools are providing AI training for teachers](https://www.edweek.org/technology/more-schools-are-providing-ai-training-for-teachers-is-it-any-good/2026/05)
[7] [University of Leicester: Microsoft collaboration](https://le.ac.uk/news/2026/june/microsoft-collaboration-university-leicester-ai-education)
[8] [Taiwan Ministry of Education: Taiwan–Korea smart-education teacher preparation exchange](https://depart.moe.edu.tw/ed2500/News_Content.aspx?n=79D4CEEC271FCB57&sms=70BE182A03B427A5&s=C6C577D8E8DEEA82)
[9] [NUS AICET: Pedagogy First—Teaching With, Not By, AI](https://aicet.comp.nus.edu.sg/aicet-learning-journey-pedagogy-first-teaching-with-not-by-ai/)
[10] [Gallup: AI use is routine among college students](https://news.gallup.com/poll/704090/routine-college-students-despite-campus-limits.aspx)
As of September 22, the most consequential global signal in AI and education is a shift from the question of whether to introduce a tool toward questions of who may use it, under what principles, and with what support. UNESCO held Digital Learning Week 2026 at its Paris headquarters from September 8 to 11. A ministerial statement presented on September 8 called for deliberative governance, public accountability, and respect for the rights of learners and teachers in the use of AI in education. UNESCO says that more than 25 education ministers and designated representatives adopted a joint statement.[1][2] It is an important cross-border policy signal, not a legally binding treaty.
UNESCO is also running a global discussion and consultation on education in the age of AI from September through November, with comments due by October 15.[3] For schools, tutoring providers, and higher-education institutions, that timetable is a useful prompt to organize evidence from practice: which uses support learning, where safeguards are needed, and where students and teachers need clearer rights and responsibilities.
## International developments
China’s Ministry of Education said on September 4 that it would implement an “AI + Teacher Development” initiative and launch comprehensive AI-literacy training for teachers through the Teacher Center of the National Smart Education Platform.[4] The public material does not specify participant numbers or a completion deadline. The careful reading, therefore, is that this is an announced direction for capacity building—not evidence that training of all teachers has already been completed.
For boundaries around student use, a September 9 Education Week report said that New York City’s school system restricted elementary- and middle-school students’ use of generative AI, while the Los Angeles Unified School District announced a restriction covering all students.[5] Separately, a nationally representative EdWeek Research Center survey of 651 teachers, conducted in February and March 2026, found that 42% reported no training on classroom use of generative AI, 22% reported multiple training sessions, and 9% reported ongoing training.[6] These are teacher-survey findings, not district-level rates or estimates for every K–12 teacher.
Several related items are best read conservatively so that announcements and collaboration discussions are not mistaken for completed implementation. On June 3, the University of Leicester announced a collaboration with Microsoft to provide full access to Microsoft 365 Copilot for all students and staff. The university materials cited here do not confirm a specific September activation date or an official headcount for those covered.[7] Taiwan’s Ministry of Education said that Seoul National University of Education President Shin-Ho Jang and a delegation visited Taiwan from September 8 to 11 to discuss digital technology and smart-education teacher preparation. AI-competency certification for student teachers was described as an area with room for cooperation, rather than a jointly completed or formally launched certification framework.[8]
At the National University of Singapore, the Artificial Intelligence Centre for Educational Technologies (AICET) is hosted by the NUS School of Computing. Its January 7 event, “Pedagogy First—Teaching With, Not By, AI,” records participation by more than 240 educators and stakeholders.[9] That is a useful pedagogical reference point, not a new September policy. In the United States, a 2026 Gallup–Lumina Foundation study found that 57% of 3,801 surveyed students pursuing associate or bachelor’s degrees said they used AI in coursework at least weekly. The survey was fielded in October 2025 among U.S. students aged 18–59 using a Dynata nonprobability web sample, so it should not be generalized to all college students or all forms of AI use.[10]
## Three questions for education institutions
**First, can governance be expressed as executable use cases?** A simple “allowed” or “banned” rule is rarely enough. Institutions can distinguish among brainstorming, practice, information organization, assessment, academic integrity, and sensitive data. For each context, they can state what students, teachers, and administrators may do, must disclose, or must not do. This approach is more realistic than treating a restriction as a total ban, while keeping rights, accountability, and teaching needs in view.
**Second, is teacher development being treated as sustained support rather than one-off tool training?** China’s announced direction and the EdWeek findings both underline that teacher capability is central to implementation. Training may include prompting, but it should also address learning goals, assessment design, bias and privacy recognition, interpretation of student work, and decisions about when not to use AI. Feedback and revision loops matter as well; completion of one workshop is not the same as durable instructional capability.
**Third, does the adoption narrative separate evidence from aspiration?** External communication should distinguish among an announcement, a pilot, access being made available, and actual activation or training completion. Likewise, room for cooperation after an international visit is not the same as a shared standard in operation, and attendance at one event is not proof of system-wide impact. Periodic review using usage evidence, teacher feedback, learning quality, and equity indicators can keep the conversation anchored in verifiable educational value.
## Conclusion
The common thread this week is that maturity in AI education will not be measured only by tool availability. It will also depend on whether institutions can define rights, responsibilities, capability support, and review mechanisms clearly. For cramschool.ai readers, the practical next step is not to chase a single platform or claim compliance with procurement requirements, certifications, or international standards. It is to build explainable, adaptable local governance and professional-learning processes that serve teaching and learning goals.
## Sources
[1] [UNESCO Digital Learning Week 2026](https://www.unesco.org/en/weeks/digital-learning)
[2] [UNESCO: Education ministers call for education to remain a common good in the age of AI](https://www.unesco.org/en/articles/education-ministers-call-education-remain-common-good-age-ai-unescos-digital-learning-week)
[3] [UNESCO global consultation on education in the age of AI](https://www.unesco.org/en/digital-education/artificial-intelligence/consultation)
[4] [China Ministry of Education: AI + Teacher Development](http://www.moe.gov.cn/fbh/live/2026/78355/mtbd/202609/t20260907_1449538.html)
[5] [Education Week: Can school districts really ban AI?](https://www.edweek.org/technology/ai-is-working-its-way-into-everything-can-school-districts-really-ban-it/2026/09)
[6] [EdWeek Research Center: More schools are providing AI training for teachers](https://www.edweek.org/technology/more-schools-are-providing-ai-training-for-teachers-is-it-any-good/2026/05)
[7] [University of Leicester: Microsoft collaboration](https://le.ac.uk/news/2026/june/microsoft-collaboration-university-leicester-ai-education)
[8] [Taiwan Ministry of Education: Taiwan–Korea smart-education teacher preparation exchange](https://depart.moe.edu.tw/ed2500/News_Content.aspx?n=79D4CEEC271FCB57&sms=70BE182A03B427A5&s=C6C577D8E8DEEA82)
[9] [NUS AICET: Pedagogy First—Teaching With, Not By, AI](https://aicet.comp.nus.edu.sg/aicet-learning-journey-pedagogy-first-teaching-with-not-by-ai/)
[10] [Gallup: AI use is routine among college students](https://news.gallup.com/poll/704090/routine-college-students-despite-campus-limits.aspx)