79% of students use AI for homework, but only 47% were shown how

OpenAI has published a report on what it calls the AI guidance gap in education. It cites new ThinkYoung research: 79% of surveyed students use AI for homework or assignments, but only 47% say they were shown how.

The announcement is dated November 10, 2026. It is signed by Tom Duff Gordon, VP of Global Affairs EMEA, and Leah Belsky, VP of Education. The report draws on OpenAI’s education partnerships with governments in Europe, the Middle East and Africa.

Two numbers frame the guidance gap

The 79% and 47% figures come from ThinkYoung’s survey of how teachers and students already use AI. OpenAI’s page does not give the sample size, the countries covered, the age range or the survey dates. Those details are presumably in the full report, which the page links to but which we have not seen.

OpenAI’s conclusion is that governments should support educators and give students clearer expectations. It says teachers need help setting the terms of use. It also argues that as AI grows more capable, teachers matter more, not less, and it points to recent debates about PISA and AI, including comments from the OECD’s Andreas Schleicher.

The partnerships behind the report cover Estonia, Slovakia, Greece and Italy, among others. OpenAI says they share three priorities: equitable access to frontier AI tools, AI literacy programmes designed jointly with the governments, and continuing measurement and evaluation. All of it is adapted to local languages, curricula and institutions.

Five themes, with evidence still thin

The report organizes its lessons under five themes: clear educational goals, educator preparation, local partnerships, critical AI literacy and credible evidence. From the Estonian, Greek and Slovak experience, OpenAI draws three lessons. They are strong public leadership that sets a direction, investment in teachers’ skills and confidence, and support tailored to local curricula and classroom needs.

The page is open about what it can show. The signals it reports are teacher adoption and reported time savings, which it calls encouraging. Whether AI strengthens knowledge, independent thinking and retention, it says, needs sustained evaluation. OpenAI describes the cases as early experiences rather than a single model to copy, and says country findings should be read within their own contexts.

That leaves the central question open. Time saved by teachers is an input measure. The page offers no results on what students learn. OpenAI says it is bringing education researchers together through the OpenAI Learning Lab to study when and how AI helps, and that it will share findings even when they challenge its assumptions.

Government deals raise privacy and oversight questions

The report presents these partnerships as policy lessons, but they are also commercial and political arrangements, and some have drawn scrutiny. Public details are limited, and the page does not describe contract terms, pricing or data handling.

Estonia was the first. In February 2025 OpenAI said it would give secondary students and teachers access to ChatGPT Edu, a version of ChatGPT built for education systems. According to The Journal, the first phase was meant to cover about 20,000 students in grades 10 and 11 and 3,000 of their teachers from September 1. Estonian reports say the AI Leap programme is funded half by the state and half by the private sector.

Greece signed a memorandum of understanding with OpenAI in September 2025, according to tech.eu and Reuters coverage. It began with a pilot that starts with upper-secondary teachers. Reporting says the contracts bar selling or sharing school data and keep it out of model training. Critics quoted in the coverage say the memorandum is not legally binding. They also point to the lack of published impact assessments and to concerns about US authorities’ access to data. A search of the coverage also found that the opposition party PASOK filed a parliamentary question on the agreement and its data-protection assessment in April 2026. We could not confirm the government’s reply.

Against that background, a survey that finds students using AI mostly without instruction is a useful argument for governments to act. It is also an argument for adopting a vendor’s tools. OpenAI makes the survey case itself, and ThinkYoung’s methodology is not described on the page, so the claim is not independently verified.

What readers can use now

The announcement is a report and a position, not a product launch, so there is nothing to switch on. It does not name new countries, give prices, or announce new plans. Its practical value is as a checklist for education ministries, schools and parents: set goals first, train teachers before students, adapt tools to local curricula, teach critical AI literacy and measure results. For students, the 32-point difference between those who use AI and those who were shown how is the finding to watch.