In November 2025, the University of Hamburg (UHH) launched a comprehensive Guidance framework for the use of generative AI systems in learning and teaching. This makes it one of those universities in Germany that set clear guidelines for the use of tools such as ChatGPT, Claude, etc. at an early stage and at the same time explore opportunities for modern, science-oriented teaching.

The new framework should provide orientation without slowing down innovation. It is published annually by Advisory Group for Digitalization in Teaching to keep up with the rapid development of generative AI.

In this article, we summarize the most important content and show why this step is not only necessary but also quit innovative in academia.

What does UHH actually understand by “generative AI”?

The framework defines generative AI (gAI) as digital systems that are based on machine learning and can generate content independently from large amounts of data such as texts, images, audio, videos or code.

Central to this is:

• AI works on the basis of statistical patterns, not on uniform factual knowledge.
• Results are not reproducible, i.e. the same prompt produces different results.
• Newer systems also combine model knowledge with current or subject-specific knowledge (RAG, retrieval-augmented generation).

The UHH is thus emphasizing: Generative AI is powerful but not reliable in a scientific sense. This is exactly where the regulatory framework comes in.

Why does the university need guidelines anyway?

Generative AI is not a trend, but a key technology that is permanently changing academic work. For UHH, the following is certain:

• AI is becoming part of almost all scientific processes.
• Students must learn to deal with them responsibly.
• Teachers need guidance on how to use AI in a meaningful and legally secure way.
• Issues such as sustainability, CO₂ consumption and ethical issues cannot be ignored.

In short: The university sees itself as having a duty to actively shape change and not just to react.

Basic didactic principles: Using AI but maintaining competencies

The framework makes it clear that AI should not replace what students should actually learn. To ensure that there is no loss of competence (“deskilling”), the UHH sets the following guidelines:

In teaching, the following applies

• AI deployment must fit to Learning objectives.
• Dealing with AI should be reflected , even if it is not actively used in the course.
• The use must be transparent.

For teachers, this means: They should try out AI, develop scenarios, clearly communicate operational limits and critically examine generative AI themselves.

For students, this means that AI can make writing processes easier, provide ideas or help with structuring, but it always entails the risk of abbreviating their own learning processes.

AI in exams: Where allowed, where prohibited and how to document?

One of the key areas of the guidance framework concerns audits. The UHH provides clarity here:

1. Exams where AI is allowed

Especially with homework or take-home exams AI should be approved in principle, as its use can hardly be proven in a legally binding manner.

BUT:
The output must documented and marked.

The framework sets out specific requirements:

• AI-generated text passages must be marked as such.
• Tools used must be listed with purpose.
• AI-based search aids must also be documented.

2. Exams where AI is prohibited

Where technically necessary, use may be prohibited. For these cases, UHH recommends alternative forms of examination:

• Focus on process documentation instead of pure product evaluation
• Oral exams or personal interviews
• Tasks that can't be solved with AI
• Presence tests under supervision

3. Declarations of independence

Students must also confirm that they have worked independently, including mention of the permitted tools. AI is automatically included, but needs to be defined more precisely.

Copyright: Who owns AI output?

Another important part of regulation: dealing with intellectual property.

Key points:

• AI-generated content is usually in the public domain because they lack “personal spiritual creation.”
• Only when users are creatively designing, a separate copyright is created.
• Prompts themselves can be protected by copyright if they are sufficiently creative.
• Teachers are not allowed to simply enter students' work into AI tools, as these are protected by copyright.

The UHH expressly recommends the use of plagiarism scanners (not AI scanners) to check unintended takeovers.

Data protection: Strict rules for personal data

The framework highlights the sensitivity of personal data:

• Such data may be used with AI systems.
• UHHGPT (the university's internal AI system) may also currently not be used with personal input.
• The legal situation regarding OpenAI & Co. remains unclear, particularly among US providers.

Only GAI systems with officially provided UHH licenses may be used compulsorily.

Particularly relevant:
US tools without certification in accordance with EU—U.S. Data Privacy Framework may no personal data received.

And what does all this mean for the future of higher education?

With this framework, the UHH is making it clear that AI will not disappear and will be an integral part of academic education but under clearly defined conditions. The university shows responsibility by:

• Innovation enabled but controlled
• Exam formats further developed
• Prepared students and teachers for a thoughtful, competent approach
• Actively takes legal, ethical and ecological aspects into account

This marks the beginning of a new phase for students and teachers. However, AI is of course not unregulated and thus leaves the long-prevailing grey area when used in a scientific context.

Conclusion

Hamburg University's new framework is a highly relevant document for dealing with generative AI at universities. It accepts the realities of life of all people working in research & teaching, creates clarity, protects scientific standards and at the same time opens up space for didactic and technical innovation.

The message is:
AI is part of the future of teaching. Nonetheless, in a responsible, transparent and scientific way.