OECD urges stronger governance as GenAI transforms higher education

The OECD has called for a more responsible and systematic approach to generative AI (GenAI) in higher education, warning that institutional policies, governance and support have failed to keep pace with the technology’s rapid adoption.

Its latest Education Spotlight finds that GenAI has moved from novelty to near-universal use among students. In the UK, undergraduate use rose from 66% in early 2024 to 95% in 2026, while comparable surveys found adoption above 90% among German students.

Across the EU, an average of 72% of students reported using GenAI during the previous three months in 2025, including 53% for formal education, although adoption ranged from almost 90% in Estonia to just over half in Türkiye.

Academic staff are also adopting GenAI, although at a slower pace. Globally, 61% of academics across 28 countries reported using it weekly, but only 17% considered themselves advanced or expert users. Most use remains focused on routine tasks such as drafting, editing, summarising and preparing teaching materials.

Adoption also varies by discipline, with science, engineering and business generally reporting greater confidence and use than the arts and humanities, while differences remain across gender, socio-economic background and institutional resources.

The diagram on page 4 maps GenAI use across teaching, learning, research, administration, student services and institutional management. Applications range from personalised learning materials and research coding to student support, regulatory compliance and large-scale analysis of internal documents.

The OECD warns that institutional governance has not kept pace with widespread adoption. Many students and staff continue to rely on free consumer AI tools because universities have yet to provide secure institutional alternatives. In the UK, only 38% of institutions actively provided GenAI tools to students in 2026, although this represented a significant increase from 9% two years earlier.

Governance is also lagging. A UNESCO survey found that only 19% of participating institutions had adopted formal AI policies, while a further 42% were still developing them. According to the OECD, this leaves students and staff to make individual decisions on issues such as privacy, copyright, ethics and appropriate academic use.

To support responsible adoption, the OECD identifies five priority policy areas: guidance for responsible use, coordinated compliance and procurement, AI skills development, evidence gathering through evaluation, and support for specialised educational tools.

Examples include national guidance in Australia, Ireland and Finland; shared procurement in France and the Netherlands; staff training in Germany, South Korea and Switzerland; and controlled pilot programmes to evaluate educational outcomes.

The report recommends guaranteeing access to devices, connectivity and secure AI tools, investing in staff development, enforcing privacy and safety standards, supporting research and maintaining meaningful human oversight alongside non-AI alternatives.

Why does it matter?

The report suggests that higher education has entered a new phase in which widespread GenAI use is no longer the primary challenge. Instead, universities and governments must determine how to integrate the technology in ways that strengthen learning while protecting academic integrity, privacy and equity.

The OECD also argues that effective AI adoption requires institutional governance rather than individual experimentation. Secure platforms, clear policies, staff training and evidence-based evaluation will increasingly determine whether GenAI enhances education or simply accelerates existing inequalities and weakens critical thinking.

Would you like to learn more about AI, tech, and digital diplomacy? If so, ask our Diplo chatbot!