Denmark has announced a package of measures to tackle the growing use of artificial intelligence in school assessments, requiring upper secondary students to orally defend their written work and subjecting computer use during exams to tighter monitoring. The steps, announced by education minister Magnus Heunicke, took effect immediately, ahead of the new school term beginning next week. "Unfortunately, we have a problem with AI cheating in upper secondary schools," Heunicke said. "Action is needed now."

The centrepiece of the new rules applies to Denmark's gymnasiet — the upper secondary school system for students aged 16 to 19, roughly equivalent to the final years of high school. Each year, around 9,000 students complete a major written assignment known as the større skriftlige opgave, or SSO, typically around 9,000 words long and carrying significant weight in their final grade. Under the new policy, students must now defend this work in a face-to-face oral examination, making it far harder to pass off AI-generated text as their own. Schools will be consulted on how best to implement the format.

Two further measures accompany the oral defence requirement. Schools must install firewall tools to block access to AI platforms during written exams and monitor students' computer screens in real time. The ministry is also urging schools to shift more written assignments into the classroom under supervised conditions, allowing teachers to observe the writing process, give timely feedback, and produce a more reliable basis for assessment.

Heunicke stressed that the three measures are only a first step, with a comprehensive national strategy for AI use across the entire education sector now in preparation. He said he would consult schools, teachers, and students to "ensure that AI does not undermine students' competencies, professionalism and, not least, the ability to think for themselves."

The challenge is far from unique to Denmark. In neighbouring Sweden, the number of students suspended or formally warned for unauthorised AI use rose by 688% between 2023 and 2025, prompting the government to commission its own review, due to report in April 2026. Across Europe and beyond, education authorities are grappling with the same fundamental tension: how to embrace AI as a legitimate learning tool while preventing it from hollowing out the skills that formal qualifications are meant to certify.