More Than 25 Countries Sign UNESCO's AI-in-Education Statement, But Guidance Still Lags Behind Classroom Use

On 8 September 2026, education ministers and designated representatives from more than 25 countries signed a joint statement at UNESCO's Digital Learning Week 2026 in Paris, committing to keep education "a human right and a common good" as artificial intelligence spreads through classrooms and administration. The statement was the political centrepiece of a four-day event that drew more than 1,000 participants in person, plus thousands online, under the theme "Education in the age of AI: Facts, Frictions, Frontiers."

The signing followed months of preparation ahead of a four-day event that, as UNESCO framed it, examined how generative and now "agentic" AI systems have moved into grading, lesson planning and tutoring since ChatGPT's public release in late 2022. Unlike a treaty or a funding instrument, the statement carries no enforcement mechanism; it is a set of principles that signatory governments agree to be judged against, not a binding commitment with penalties for departure.

Among the eight commitments reported from the text: AI systems used in education should remain open to public review, students should be helped to build critical thinking rather than defer to AI outputs, teachers should have a say in which tools enter their own classrooms, and families and schools should retain the ability to pause or stop AI use. The statement also asks governments to treat learners' and teachers' data rights as a governance priority, alongside age-appropriate design for AI tools used with children.

The guidance gap UNESCO has tracked since 2023

The statement did not appear in a vacuum. UNESCO has been measuring the distance between AI adoption and AI policy in education for three years, and the gap it has documented is what gives the Paris statement its urgency. In a global survey conducted in May 2023 of just over 450 schools and universities, UNESCO found that fewer than 10% had developed institutional policies or formal guidance on generative AI. Roughly 40% of the institutions that did have guidance had only communicated it orally, and close to one in five respondents said they simply did not know whether their institution had a policy at all.

By September 2025, a follow-up survey of 400 UNESCO Chairs and UNITWIN network institutions across 90 countries found the picture had shifted: 19% now had a formal AI policy in place and a further 42% were drafting one, putting roughly two-thirds of respondents on a path toward institutional guidance. That progress was uneven by region — around 70% of institutions in Europe and North America had guidance or were developing it, compared with 45% in Latin America and the Caribbean.

Measure	UNESCO survey, May 2023	UNESCO survey, Sept. 2025
Institutions with formal AI guidance	<10% (450+ schools/universities)	19% formal, 42% in progress (400 respondents, 90 countries)
Europe & North America	44% of respondents (regional share)	~70% with or developing guidance
Latin America & Caribbean	17% of respondents (regional share)	45% with or developing guidance

Policy has moved, in other words, but from a very low base, and mostly in the two regions best represented in UNESCO's own survey samples. Whether that pattern holds where response rates were thinner is not something either survey can settle on its own.

Independent data: use has outrun policy almost everywhere

A separate, independent measure released the same year points to a wider gap between use and governance than either UNESCO survey alone would suggest. The Digital Education Council's AI in Higher Education Global Survey 2026, published on 9 July 2026 from 45,398 responses across 35 countries, found that 88% of students and 77% of faculty now use AI in their learning or teaching — faculty use up 16 percentage points on the previous year's edition of the same survey, according to reporting on the findings by EdTech Innovation Hub. Regional intent to keep using AI diverged sharply: faculty intent rose to 92% in Asia-Pacific, 89% in Europe/Middle East/Africa and 94% in Latin America, while it fell nine points in the United States and Canada, from 76% to 67%.

The same survey found signs of strain alongside the growth: 57% of students said their assessments offered inadequate guidance on AI use, only 28% felt their assessments reflected the skills an AI-enabled workplace would require, and about one in five reported retaining less information or losing confidence in unassisted work because of routine AI reliance. In the United States and Canada, 81% of students said AI discourages critical thinking, against 66% globally, and 43% said they would support an outright institutional ban — a level of skepticism not matched in any other region surveyed.

What the statement does not settle

None of this is mentioned in the Paris statement itself, which sets out aims rather than obligations. UNESCO paired the signing with a compilation of policy briefs on AI safety governance, digital sovereignty, procurement and cost models, intended to give ministries a starting template rather than a rulebook.

As independent previews of the event noted, questions about synthetic research outputs and AI-generated "ghost learners" were framed at Digital Learning Week as open problems, not settled ones. Whether the commitments signed on 8 September translate into the kind of enforceable national guidance that was still missing from most institutions as recently as 2023 will depend on decisions each of the more than 25 signatory governments makes on its own, over the next several budget and legislative cycles.