Main Content
AI-Powered Services (GWDG AI Cloud)
The information below is based on the linked guidelines (PDF) (GER), which must be strictly followed when using the service.
Marburg University offers all members of the university the opportunity to use generative AI free of charge and in compliance with data protection regulations. The following services are provided in collaboration with the AI Service Center for Sensitive and Critical Infrastructure (KISSKI) of the Gesellschaft für wissenschaftliche Datenverarbeitung mbH Göttingen (GWDG) via the so-called Academic Cloud.
The various AI applications are being used as part of a trial run. This means, among other things, that settings, functions, or operating conditions can be dynamically adjusted, which may also result in a loss of functionality.
These various web-based platforms can be used to perform a wide range of tasks, such as text generation, translation, and image generation.
Chat-AI: Text-Generating AI Language Models
The Chat-AI service provides access to various text-generating AI language models. Among other things, these enable the generation, analysis, and translation of text. Both open-source models and commercial models, such as OpenAI ChatGPT, are available.
Voice-AI: AI-powered speech processing
The Voice-AI service is based on Whisper, a powerful open-source language model for automatic speech recognition. It enables the analysis, transcription, and translation of audio recordings.
Image-AI: AI-powered image generation
Image-AI is based on FLUX, a powerful open-source language model for image generation. It enables AI-powered image generation based on text input (text-to-image).
Important Rules for Use
- Use of the service is voluntary. In particular, employees must not suffer any disadvantages as a result of not using the service.
- Never enter personal, confidential, or security-related data via the prompt, and never provide such data to the AI model via files.
- Commercial use of the service is not permitted.
- If the service is used, this must be clearly stated.
- Do not accept generated content without verifying it. It may be incorrect, misleading, incomplete, unreliable, or biased. Even for content generated by chat AI, you are responsible for evaluating and ensuring the accuracy of your work results.
- Under no circumstances should you use the service for cognitive behavioral manipulation, social evaluation, biometric identification and categorization of individuals, automated grading of exams, or other forms of automated performance and aptitude assessment, such as in the context of personnel selection. Research and teaching projects in these areas must comply with all relevant scientific, ethical, and legal requirements.
- Take advantage of the training and support services offered by Marburg University on the topic of generative AI (see the "Training" section).
Evaluation and Feedback
During the pilot phase, users are invited to provide feedback and suggestions for improvement. Opportunities to do so will be available at central events or through surveys. Users are encouraged to informally discuss their experiences with the tool within their own work and study environments and to reflect on their own use of generative AI.
Benefits for You
The goal of this initiative is to give university staff and faculty the opportunity to experiment with generative AI in research and teaching, potentially integrate it into their own workflows, and engage in internal discussions about the use of AI. This initiative lays another foundation for the thoughtful, responsible, and forward-looking use of new technologies at the university.
- Support for research and teaching: AI can assist with structuring content, generating new ideas, or translating academic texts
- Free and GDPR-compliant access to various text-generating AI chat models for Marburg University staff
- Access via AcademicCloud using your university account (Single Sign-On, SSO)
- Focus on data protection: Unlike commercial AI services, this service provides a secure environment within the university’s infrastructure
Target Audience
Faculty, researchers, students, staff
Requirements
- Central user account (university account)
- Up-to-date web browser
- Before using AI services or models from the GWDG AI Cloud, users must complete the introductory module on the use of generative AI at Marburg University and pass the associated test.
Training Courses
- Introductory module on the use of generative AI at Marburg University (GER)
- Guidance for instructors on working with language-based AI models (GER)
- Learning module on generative AI (ILIAS self-paced module) (GER)
- The AI Campus offers a wide range of learning resources on the topic of AI
Access
You can easily access the system using your Marburg University account via federated authentication:
On the login screen, select the "Förderierte Anmeldung" button on the right.
Next, select "Philipps-Universität Marburg" from the dropdown menu or search bar. Then click "Weiter" (continue) to be redirected to Marburg University's login page. Enter your login credentials as usual and accept the terms of use.
Status
In operation since July 21, 2025
Inhalt ausklappen Inhalt einklappen Support and Help
Inhalt ausklappen Inhalt einklappen Related Services
Integration with ILIAS; see "Good to Know"Inhalt ausklappen Inhalt einklappen Good to Know
Integration Capabilities
Another advantage of the GWDG AI Cloud is its seamless integration with other core IT services at Marburg University. For example, integration with existing systems such as the ILIAS learning platform has already been successfully implemented, expanding the functionalities within the teaching and learning environment.
Handling of Data
The GWDG places great importance on data protection. Your entries and responses are stored only locally during a session and are deleted when you close the website. Only the number of requests per user is recorded for the purpose of assessing demand.
Please note: This applies only to internal LLMs (Mixtral, Qwen, Meta). While GWDG does not share any user data with OpenAI, queries are sent directly to the external service provider!
For more details, please see the Privacy Policy.
Data Protection
In the event of any data protection incidents, such as the accidental entry of personal data, the Data Protection Officer at Marburg University must be notified immediately.