Copyright & AI

• A duty to label may arise from the terms of use of the software*, the relevant examination regulations, or the university's framework regulations.
  (*Of the larger AI generators, only Midjourney imposes a restriction on its free accounts: it states that Midjourney output may only be made available under CC BY-NC (no commercial use). However, with the NC restriction, the output can no longer considered as OER. (Rack 2023))
• The University of Basel's guide to citing AI (Leitfaden Aus KI zitieren) provides advice on citing and documenting content created using artificial intelligence-based tools.
• Depending on the citation style, the citation varies. The 18th edition of the Chicago Manual of Style already explicitly outlines rules for dealing with AI-generated content. For other citation styles, the governing bodies and libraries have addressed the topic of AI (specifically ChatGPT) in blogs and website posts: APA, MLA, IEEE.
• Consult the relevant guidelines for academic work at your professorship to ensure that you correctly cite the use of AI-supported tools. Here are some examples from the design guidelines for academic work published by the Centre of Techer Training and Educational Research of Chemnitz University of Technology:
  • Direct labeling of an image
    • Figure 1: A large stack of books, image generated by Stable Diffusion, February 28, 2024
  • Mention as a tool
    • Stable Diffusion, Stability AI: https://stablediffusionweb.com A large stack of books, image generated on February 28, 2024
  • Translation
    • DeepL Translate, Deepl SE: https://www.deepl.com/translator Translation of text passages

Licensed e-books may not be uploaded to AI tools, as the content is protected by copyright. This use case needs to be included in future license agreements.

• Content with a CC license can be used freely in accordance with the license terms and may allow more. (Kreutzer 2025)
• Apart from copyright regulations, other requirements must also be taken into account: copyright of the AI application, right of personality, data protection, ethical issues, recognizability of protected works by third parties. (Brehm 2024)

• Depending on the type of use, the output may be permitted, restricted, or subject to compensation. (Brehm 2022, S. 12)
• AI can't be the creator of a new work. A new work is only something that's a unique intellectual creation (§2 para. 2 UrhG). But if AI is just used as a tool and a human is mainly responsible for the creative input, then it can be a new work that's protected by copyright.
• However, if a human dictates all essential design decisions through a detailed prompt or chain of strongly steered prompts, and the AI program appears only as an executing instrument, or if users further work with the generated text as food for thought, it can be assumed that the work is protected by copyright. (Salden 2023)

• Yes, a CC license does not prevent the use of content for AI training (Kreutzer 2021), not even the CC BY-NC license.
• Unlike § 60d of the German Copyright Act (UrhG), TDM may be prohibited by the copyright holder through a reservation of use pursuant to § 44b (3). This also applies to licensed databases whose contents are not protected by copyright or whose publications are made available under open access licenses. The use of individual publications is possible, but not the use of substantial parts (database law). Legal restrictions, e.g., fair use (USA) or TDM-restrictions (EU), take precedence over the CC license. Detailed information on the individual CC licenses and their use for AI can be found in Kreutzer 2025.
• In the context of CC licenses, the use of articles for training AI tools could be regulated in the future via so-called preference signaling.
• The use of content licensed under a CC BY-ND license is also permitted to others, for example, for data processing, creating a text corpus, and archiving. However, this is limited to internal use, e.g., as part of a research project at a university. (Brehm 2024).

Provided that no intellectual property rights arise, the content can be used in its entirety for OPAL. If intellectual property rights apply, UrhG §60 c applies and 15% of the work can be used in OPAL.

Yes, this is permitted. AI-generated text that is created without significant human influence is considered to be in the public domain under copyright law. (Salden 2023). If users of AI software can assert copyright for AI-generated text, licensing as OER is possible. However, it must be ensured that the AI-generated text does not contain any copyright-protected content.

Uploading a student's work to an AI detector without consent to check whether it was written using AI is problematic from a copyright perspective. (Salden 2023) Exam results are protected by copyright. This data must not be entered into AI software if it is to be reused as training data or used for other purposes. The assessment of the work itself can only be carried out by the examiner and not by software. AI systems may be excluded as assistant tools in examination regulations or restricted to certain applications. (Bischof 2025, Präsidium DFG 2023)

• to help design teaching materials,
• as a learning aid for students,
• as a tool for translating teaching materials,
• to generate exercises and coursework assignments.

Legal restrictions and DFG guidelines have to be followed.

Brehm, E. (2022) Guidelines zum Text und Data Mining für Forschungszwecke in Deutschland. TIB – Leibniz-Informationszentrum Technik und Naturwissenschaften und Universitätsbibliothek. Available at: https://oa.tib.eu/renate/bitstream/123456789/10352/3/GuidelinesTDM_NFDI4Ing_oeff.pdf (Accessed: 27 May 2025).

Brehm, E. (2024) Open Access und Künstliche Intelligenz. Was ist erlaubt und was nicht?, Vortragsfolien. Available at: https://open-access.network/fileadmin/ipoa/Veranstaltungen/oa-talk/oatalk_20240926_brehm.pdf (Accessed: 27 May 2025).

Bundesministerium für Bildung und Forschung – BMBF (2025) Künstliche Intelligenz. Available at: https://www.bmbf.de/DE/Forschung/Zukunftstechnologien/KuenstlicheIntelligenz/kuenstlicheintelligenz_node.html (Accessed: 27 May 2025).

Creative Commons (2024) Six Insights on Preference Signals for AI Training, Blogbeitrag. Available at: https://creativecommons.org/2024/08/23/six-insights-on-preference-signals-for-ai-training/ (Accessed: 27 May 2025).

Creative Commons (2025) Using CC-licensed Works for AI Training. Available at: https://creativecommons.org/wp-content/uploads/2025/05/Using-CC-licensed-Works-for-AI-Training.pdf (Accessed: 27 May 2025).

Kreutzer, T. (2021) Daten und Creative-Commons-Lizenzen - Trainingsmaterial für Künstliche Intelligenz, Blogbeitrag. Available at: https://irights.info/artikel/daten-und-creative-commons-lizenzen-trainingsmaterial-fuer-kuenstliche-intelligenz/31086 (Accessed: 27 May 2025).

Kreutzer, T. (2024) Open content - navigating creative commons licenses. Bonn: Deutsche UNESCO-Kommission e.V. Available at: https://irights.info/wp-content/uploads/2025/01/Open-Content_final_web.pdf (Accessed: 27 May 2025).

McAdoo, T. (2023) How to cite ChatGPT, Blogbeitrag. Available at: https://apastyle.apa.org/blog/how-to-cite-chatgpt (Accessed: 27 May 2025).

MLA (2023) How do I cite generative AI in MLA style? Blogbeitrag. Available at: https://style.mla.org/citing-generative-ai/ (Accessed: 27 May 2025).

Nanyang Technological University Library (2025) LibGuides: Citation Styles: How to cite Gen AI content in IEEE Style. Available at: https://libguides.ntu.edu.sg/c.php?g=935202&p=7006284 (Accessed: 27 May 2025).

OER FAQ (no date) Was ist die Public Domain Mark?. Available at: https://oer-faq.de/faq/was-ist-die-public-domain-mark/ (Accessed: 27 May 2025).

Präsidium der Deutschen Forschungsgemeinschaft (2023) Stellungnahme des Präsidiums der Deutschen Forschungsgemeinschaft (DFG) zum Einfluss generativer Modelle für die Text- und Bilderstellung auf die Wissenschaften und das Förderhandeln der DFG. Available at: https://www.tu-chemnitz.de/urz/apps/ki/230921_Stellungnahme_Praesidium_DFG_ki_ai.pdf (Accessed: 27 May 2025).

Rack, F. (2023) OER und CC-Lizenzen bei generativer KI, Blogbeitrag. Available at: https://irights.info/artikel/oer-cc-lizenzen-generative-ki/32090 (Accessed: 27 May 2025).

Salden, Peter and Leschke, J. (2023) Didaktische und rechtliche Perspektiven auf KI-gestütztes Schreiben in der Hochschulbildung. Available at: https://doi.org/10.13154/294-9734.

Universitätsrechenzentrum der TU Chemnitz (2025) KI in der Hochschullehre. Available at: https://www.tu-chemnitz.de/urz/apps/ki/ (Accessed: 27 May 2025).

Universität Basel (2024) Leitfaden «Aus KI zitieren». Available at: https://www.unibas.ch/dam/jcr:e46db904-bf0f-475a-98bc-94ef4d16ad2e/Leitfaden-KI-zitieren_v2.2.pdf (Accessed: 27 May 2025).

University of Chicago Press (ed.) (2024) The Chicago manual of style: the essential guide for writers, editors, and publishers, 18th edition. Chicago: The University of Chicago Press.

Bischof, A. (2025) "ChatGPT & generative KI: Handreichung für den Umgang in Lehre und Studium an der Fakultät für Human- und Sozialwissenschaften." Available at: https://www.tu-chemnitz.de/hsw/soziologie/Professuren/Techniksoziologie/Lehre/HSW_Handreichung_Taskforce_ChatGPT.pdf (Accessed: 27 May 2025)