Denmark’s education minister has launched an emergency “AI package” for high schools as teachers scramble to deal with ChatGPT in exams. Gymnasium organizations call it a necessary first step but warn the real work still lies ahead.
The Danish Ministry of Education has unveiled what it calls an AI strakspakke, an emergency set of guidelines aimed at helping gymnasiums handle artificial intelligence in classrooms and exams. The move comes as gymnasium organizations representing school leaders, teachers, and students publicly acknowledge the problem has reached crisis levels.
The package includes updated ministry guidance on AI use in teaching and exams, tighter requirements for school conduct rules, and recommendations to redesign assignments so they are harder to fake with generative AI. Schools are also being urged to increase oral exam components, require process documentation, and use synopses to track student work over time.
Teachers Say Trust Is Breaking Down
I have watched Danish gymnasiums struggle with this for two years now. Since ChatGPT arrived in late 2022, teachers tell me they can no longer be sure who wrote what. A Danish essay that once took hours of real work can now be generated in seconds. Traditional homework formats, especially written reports and project work, have become dangerously vulnerable.
The three major gymnasium organizations, Danske Gymnasier, Gymnasielærerne, and Danske Gymnasieelevers Sammenslutning, issued a joint statement welcoming the package. But their praise was measured. They describe the measures as solving some of the most pressing short term challenges while emphasizing that the real structural problems remain unaddressed.
What the Package Actually Does
The strakspakke is not new legislation. It is a set of administrative guidelines and recommendations that schools can implement immediately. The ministry is urging gymnasiums to monitor exams using uniform national standards. Schools are being told to increase the amount of written work done in supervised classroom settings.
One concrete proposal targets the larger written assignment known as SSO on the hf track. Students will now have to defend their work orally, a format already common on other tracks. The goal is to make it harder for AI generated text to slip through undetected.
Maja Bødtcher Hansen, chairperson of Danske Gymnasier, called it an important first step that gives schools concrete tools to strengthen trust in exams. But she also made clear that the sector is looking ahead to more fundamental solutions that will take years to develop.
The Scale of the Problem
Denmark does not have hard national statistics on how many students are using AI to cheat. That is part of the problem. Teachers report rising suspicions but proving AI use is technically difficult. Detection software is unreliable and prone to false positives, which makes it risky to use as evidence in disciplinary cases.
What is clear is that the anxiety is real and widespread. Gymnasium teachers I have spoken to describe a breakdown in the basic contract between student and teacher. When you can no longer tell if the work is genuine, grading becomes a farce. Exam results lose meaning, and the entire legitimacy of the diploma system is at risk.
A Late Response by European Standards
Denmark is playing catch up here. Norway issued national guidance on ChatGPT in education over a year ago, with clear guidelines on acceptable and unacceptable use. The Netherlands did something similar through its ministry and Kennisnet, a public education technology service.
Swedish schools have launched regional initiatives, though Sweden has not yet settled on a single national approach. Danish gymnasium organizations have pointed to these neighbors as proof that a national framework is both feasible and necessary to avoid a patchwork of conflicting school rules.
The strakspakke brings Denmark closer to the European norm but it arrives late. The delay has left individual schools to improvise their own policies, often with wildly different results. Some have banned AI outright in homework. Others encourage critical use. Students moving between schools, or even between teachers, face inconsistent expectations.
The Equity Problem No One Wants to Talk About
There is a class dimension to this that rarely gets mentioned in official statements. Students with money can access advanced paid AI tools that produce far more sophisticated output than free versions. This creates an uneven playing field where wealthier students have a clear advantage, not just in completing assignments but in evading detection.
I find it striking that this issue barely features in the ministry package or in the public statements from gymnasium organizations. It should. If AI becomes a standard part of student work, Denmark risks embedding a new form of inequality directly into its education system.
Two Competing Visions
The real conflict here is not about whether to regulate AI. Everyone agrees some regulation is needed. The disagreement is about what kind. One camp wants strict control, more supervision, and technological enforcement. The other wants to integrate AI into teaching as a tool students learn to use critically.
The control camp argues that unregulated AI use destroys exam validity and undermines the entire grading system. They point out that if students can generate high level work with minimal effort, diplomas become meaningless. They also worry about dependency, where students lose the ability to write, think critically, or evaluate sources on their own.
The integration camp, which includes many teachers and education researchers, counters that banning AI is both futile and backwards. They argue that schools have always had to adapt to new technology, from calculators to the internet. The job of education is not to resist change but to teach students how to use tools responsibly. They warn that a ban driven approach will push AI use underground, where students get no guidance at all.
What Happens Next
The strakspakke is explicitly framed as a temporary fix. The ministry has promised a broader AI strategy for gymnasiums, a new expert group, and evaluations of existing pilot projects. That work will take time, possibly years. In the meantime, schools will operate under the new guidelines starting this school year.
Anders Frikke, chairperson of Gymnasielærerne, praised the minister for acting quickly but also listed additional measures his union wants to see. These include mandatory declarations when AI is used in major written assignments, oral exams without AI access, and more sub exams conducted without digital tools.
Oscar Tønsberg Hoffmann, chairperson of the student organization DGS, emphasized that students want clear and uniform rules and exams that feel fair. He also stressed that students need to be involved in shaping the long term solutions, not just told what the rules are after the fact.
A Broader Question About Danish Education
This debate is not really about ChatGPT. It is about what Danish education is for. If the goal is to certify that students can perform specific tasks without help, then AI is a direct threat. If the goal is to prepare students for a world where AI is everywhere, then the current exam system is already obsolete.
I suspect Denmark will end up somewhere in between, as it usually does. The strakspakke buys time. It reassures parents and teachers that someone is paying attention. But it does not resolve the deeper tension between tradition and transformation. That fight is just beginning.
Sources and References
Ritzau: Gymnasieorganisationer: Ministerens AI-strakspakke er et vigtigt første skridt
The Danish Dream: AI Skills Now Essential in Danish Job Market
The Danish Dream: Danes Turn to AI Like ChatGPT for Diagnoses
The Danish Dream: Denmark Debates AI Sermon Tool for Pastors
