Albert-Ludwigs-Universitaet Freiburg

Warunki i wyjątki

Mandatory training; independent output review; high-risk approval with AI contact and, for personal data, DPO.

Referenced AI Directive pending committee process asof July2026, not certified enacted.

Diese Nutzungsbedingungen werden regelmäßig überprüft und weiterentwickelt. Änderungen werden rechtzeitig bekannt gegeben. Terms of Use for OpenWebUI for Generative AI Systems at the University of Freiburg These terms of use regulate the permissible use of the Open WebUI tool and the local and external AI models connected via this tool. The goal is to ensure a legally secure, responsible, and sustainable use of AI systems at the University of Freiburg. These terms of use concretize the "Directive on the Use and Operation of AI Systems at the Albert Ludwig University of Freiburg" (AI Directive*) for the specific system Open WebUI. The AI Directive remains unaffected and takes precedence in the event of contradictions. Open WebUI is a supportive work tool. It does not replace professional, legal, or organizational decisions and is not designed as an autonomous specialist procedure or agentic system. All users are required to comply with the specified guidelines and notices to ensure a legally secure and ethically acceptable use of Open WebUI. * The AI Directive is currently (as of July 2026) in the university's committee process and will enter into force upon its publication. I. Mandatory AI Training Before first use of Open WebUI, all users are required to complete the training course "Basics of AI: Understanding and Applying Generative Artificial Intelligence (AI)" available on the ILIAS learning platform. The training serves to acquire the required AI competence as defined by the AI Directive. Further training requirements are governed by the AI Directive. II. Principles of Using OpenWebUI These terms of use regulate the use of the Open WebUI tool provided by the University of Freiburg, particularly the handling of inputs and outputs. Input refers to all user inputs, such as texts, documents, tables, images, or audio recordings. Output refers to the results generated by the connected AI models. The responsibility for entered content and for the use of the generated results lies with the user. Results must be independently checked for factual accuracy, completeness, legal permissibility, and potential discriminatory effects. Generative AI can produce erroneous or biased content (hallucinations). Results must not be adopted without review or used as the sole basis for decision-making. The use of Open WebUI is generally permitted, provided that the legal requirements, in particular data protection, copyright, and higher education law, as well as the provisions of the AI Directive of the University of Freiburg, are complied with and no prohibited AI practices within the meaning of the AI Regulation exist. The usage scenarios documented in Annex 1 represent reviewed and approved standard applications. They serve as guidance and concretize typical use cases in research, teaching, and administration. A brief overview of the most important rules for use ("What am I allowed to do, what am I not allowed to do") can also be found in the usage scenarios (Annex 1) and in the short guide (Annex 2). Other use cases are also permitted without separate approval if they comply with the legal requirements and do not give rise to increased risk within the meaning of these terms of use. In case of doubt, prior consultation with the responsible AI contact person in accordance with § 9 para. 3 of the AI Directive is required. Use cases with increased risk A use case with increased risk exists in particular when special categories of personal data within the meaning of Art. 9 GDPR are to be processed, sensitive administrative, personnel, or application data is affected, a permanent technical integration or automation (e.g., API-based further processing) is planned, use takes place in the context of admission, examination, or personnel decisions, or image data depicting persons is processed. In these cases, approval by the responsible organizational unit with the involvement of the AI contact person is required before use; in the case of processing personal data, the involvement of the Data Protection Officer is additionally required.

Zebrano: 2026-10-08 · 75–96
SHA-256: 7040fb57b8b183b09d72f66f1d0f1f5f86694e9c9ded29cd1643f25cfb2fe17e

2. Studies and Teaching The use of Open WebUI in studies and teaching is permitted in accordance with §§ 4, 5 of the AI Directive. Specific permitted usage scenarios are regulated in Annex 1. Automated assessment of examination performance and analysis of student behavior are not permitted. Examination decisions must always be made by a responsible natural person. 3. Administration In central and decentralized university administration, Open WebUI may be used in a supportive capacity within the framework of § 6 of the AI Directive. Permitted uses include in particular: Creating, revising, or translating texts, Structuring information, Analyzing documents, Preparing decision bases, proposals, and concept papers, Drafting emails, Creating material-based chatbots, as well as the usage scenarios regulated in Annex 1. Decisions with legal effect must always be reviewed and ultimately assumed by a human entity (cf. § 6 para. 3 of the AI Directive). IV. Processing of Personal and Confidential Data The processing of personal data by Open WebUI is furthermore governed by § 7 of the AI Directive and the data protection regulations. Personal data may only be processed if a legal basis exists and the processing is limited to what is necessary. Input within the framework of the usage scenarios regulated in Annex 1 is generally permitted. In use cases with increased risk, the use must be documented appropriately. Documentation is carried out by the respective responsible organizational unit with the involvement of the AI contact person. Information on internal processes, strategies, financial and personnel matters, IT security concepts, or contractually protected content may only be processed if this is professionally necessary and legally permissible.

Zebrano: 2026-10-08 · 112–127
SHA-256: 7040fb57b8b183b09d72f66f1d0f1f5f86694e9c9ded29cd1643f25cfb2fe17e

Verantwortlichkeit (1) Die Verantwortung für KI-generierte oder beeinflusste Inhalte liegt bei den Personen, die sie erstellen oder nutzen. Dies gilt z.B. für falsche Darstellungen, Plagiate, Autorschaft und Urheberrechtsverstöße. (2) KI-generierte oder -beeinflusste Inhalte sollen von Studierenden angezeigt sowie erklärt werden können. Transparenz, Dokumentation und wissenschaftliche Integrität (1) Bei der Verwendung generativer KI gelten die Grundsätze guter wissenschaftlicher Praxis, insbesondere, weil eine flächendeckende automatisierte Identifikation von KI-generierten Inhalten weder technisch zuverlässig möglich noch epistemologisch wünschenswert ist. Hierzu wird auf die Leitlinien zur Redlichkeit in der Wissenschaft der Universität Freiburg und der DFG verwiesen 8. (2) Die Anforderung an die Dokumentation 9 von KI-Nutzung kann je nach Fachbereich und Art der Studien- bzw. Prüfungsleistung variieren. Empfohlen werden Angaben zum Verwendungszweck, zu betroffenen Inhalten und genutzten Modellen. Grundsätzlich gilt: • Rechtschreib-, Grammatik- und Stilkorrektur sowie maschinelle Übersetzungshilfen müssen gesondert angegeben werden. • KI-gestützte Recherche erfordert eine Prüfung und Korrektur von Inhalten. • Wissenschaftliche Aussagen müssen anhand überprüfbarer Quellenangaben belegt werden. • KI-generierte Inhalte sind keine zitierfähigen wissenschaftlichen Quellen.

Zebrano: 2026-10-08 · 236–260
SHA-256: fb91a2c36d8bfaeb06dccbbda4fa33ea1ed2d69aa509ec8405cdb6ccd1179829

Warum Open WebUI nutzen? Vielfältige Modellauswahl: Greifen Sie auf eine wachsende Palette interner Modelle (vom Rechenzentrum bereitgestellt) RAG – Retrieval‑Augmented Generation: Arbeiten Sie nicht nur textbasiert, sondern binden Sie eigene Dokumente, PDFs oder Datenbanken ein – das Modell liefert Antworten, die auf Ihrem Wissensbestand basieren. Individuelle Chatbots: Erstellen Sie in wenigen Klicks spezialisierte Bots für Ihre Fachbereiche, Support‑Teams oder Forschungsprojekte. Datenschutz + Compliance: Alle Daten bleiben bei Verwendung der internen Modelle innerhalb der gesicherten Uni‑Umgebung. Bei der Nutzung externer Modelle werden Ihre Anfragen über den zentralen API Zugang der Universität an die Anbieter weitergeleitet. Einfacher Zugriff: Keine Installation nötig – alles, was Sie benötigen, ist ein aktueller Browser und einen gültigen UniAccount. Wie fange ich an? Basisanleitung Open WebUI: Für einen ersten Einstieg haben wir eine Basisanleitung für Sie erstellt. Öffnen Sie https://openwebui.uni-freiburg.de in einem Browser Ihrer Wahl Login über „Mit UniAccount der Universität Freiburg fortfahren“ Modell wählen – intern, z.B. Standard Chat UFR oder extern (OpenAI, Mistral AI). Prompt eingeben oder Dokument(e) hochladen → sofortige KI‑Unterstützung. Ergebnisse nutzen – Text, Code, Zusammenfassungen, Analysen oder maßgeschneiderte Chatbot‑Dialoge. Schulungen & Datenschutz Nutzungsbedingungen für Open WebUI: Die Nutzungsbedingungen finden Sie unter https://openwebui.uni-freiburg.de/nutzungsbedingungen/ Basisschulung KI: Verstehen und Anwenden von generativer Künstlicher Intelligenz (KI): Bevor Sie Open WebUI nutzen dürfen, müssen Sie die Online-“Basisschulung KI“ auf der Lernplattform ILIAS durcharbeiten und beim erste Login die Nutzungsbedingungen akzeptieren. Das Schulungsmodul „Verstehen und Anwenden von generativer Künstlicher Intelligenz (KI)“ ist auf die Anforderungen der EU-KI-Verordnung abgestimmt und dient an der Universität Freiburg als verpflichtende Schulung für die Nutzung von generativer KI und LLMs über die universitären Werkzeuge.

Zebrano: 2026-10-08 · 24–41
SHA-256: 738efe1bcd7ad1a8139a8f7a342ce5ed07d77575d25b81d05a303aee5f971e13

Warunki i wyjątki

Lawful RAG source materials and sharing basis; approved models/projects, IT security and possible daily limits.

V. Uploading Documents and Knowledge Databases (RAG) Uploading documents to configure personal knowledge databases (Retrieval Augmented Generation – RAG) is permitted if the source materials may be used lawfully. Uploaded files are technically processed and stored in a vector database. Automatic updating does not take place. Users are responsible for the currency and lawfulness of the stored content. Cross-departmental sharing of knowledge databases may only take place if a legal basis exists for this purpose. VI. Use of External AI Models, API Usage, and Usage Volume Open WebUI provides access to multiple Large Language Models (LLMs), including those from OpenAI (ChatGPT) or mistralAI, via an API. Inputs to third-party providers are processed according to their policies and may be stored temporarily. Your inputs are not used as training data. Chat content entered in Open WebUI does not become training data for the third-party LLMs. The use of Open WebUI is limited to the models approved by the University of Freiburg. Further model restrictions may apply to certain use cases. The provisions in Annex 1 are authoritative. The use of the Open WebUI API is only permitted within the framework of approved projects, for study or teaching purposes, or as part of personal official use (e.g., programming environments) and in compliance with IT security requirements. Unauthorized connection of services that create a sustainable dependency or high automated usage load (e.g., batch processing) is not permitted. Please contact ki@rz.uni-freiburg.de for clarification. Usage volume To ensure fair and resource-efficient operation, a daily usage volume may be set for the use of Open WebUI. This may relate in particular to the number of requests, the processed token volume, or a monetary usage equivalent. If the set daily limit is reached, the following message appears: "You have reached your current daily request limit. If you need increased usage for a research project or for any other reason, please contact ki@rz.uni-freiburg.de." An adjustment of the individual usage volume is possible upon reasoned request. The decision is made within the framework of available technical and financial resources. VII. Documentation of Usage Scenarios (Annex 1) The specifically reviewed and approved usage scenarios of Open WebUI are documented in Annex 1. Annex 1 is maintained as a continuously updated document. Updates are carried out by the AI Commission. Changes and additions require approval by the Data Protection Officer and the Information Security Officer of the university. Updates are announced in Open WebUI and are binding from the time of announcement. VIII. Sustainability Guidelines Users should be aware that the provision and use of generative AI models involve high energy consumption. Requests should only be made when necessary.

Zebrano: 2026-10-08 · 128–140
SHA-256: 7040fb57b8b183b09d72f66f1d0f1f5f86694e9c9ded29cd1643f25cfb2fe17e