EU Copyright Law and the Protection of Computer Code Created with AI Assistance
(2025) JURM02 20251Department of Law
Faculty of Law
- Abstract
- This thesis examines whether computer code generated with the assistance of artificial intelligence (AI) can qualify as a “work” under EU copyright law, and if so, who may be recognized as its author. As AI programming tools such as GitHub Copilot and Tabnine become more integrated into software development, they increasingly mediate the creative process—raising important legal questions concerning originality, human intellectual effort, authorship, and enforcement.
Applying a legal dogmatic method, the thesis first establishes the EU’s harmonized criteria for copyright protection, as developed through the CJEU’s case law. To qualify as a “work”, a subject matter must fall within a recognized domain, result from human intellectual... (More) - This thesis examines whether computer code generated with the assistance of artificial intelligence (AI) can qualify as a “work” under EU copyright law, and if so, who may be recognized as its author. As AI programming tools such as GitHub Copilot and Tabnine become more integrated into software development, they increasingly mediate the creative process—raising important legal questions concerning originality, human intellectual effort, authorship, and enforcement.
Applying a legal dogmatic method, the thesis first establishes the EU’s harmonized criteria for copyright protection, as developed through the CJEU’s case law. To qualify as a “work”, a subject matter must fall within a recognized domain, result from human intellectual effort, reflect free and creative choices, and have these choices expressed in the final output. The thesis finds that AI-assisted computer code can meet these criteria where meaningful human input is present—particularly in the conception and redaction phases of the creative process. However, where user involvement is minimal, the resulting code may fail to satisfy the originality requirement.
The analysis further shows that the threshold of originality under EU law remains relatively low, provided the author retains creative discretion despite the use of technical aids. This is especially relevant for AI-generated outputs, where the human role may be indirect or fragmented. Nonetheless, originality presupposes that the human user exercises actual creative judgment; mere reliance on fully AI-generated outputs will not suffice.
In considering authorship, the thesis emphasizes that EU copyright law remains anthropocentric: only human creators may be recognized as authors. The absence of a harmonized definition of authorship at the EU level creates some legal uncertainty, particularly in the context of AI-assisted creation. Nonetheless, the CJEU’s jurisprudence has consistently linked authorship to human intellectual creation. As such, the thesis finds that current EU law does not permit the attribution of authorship to AI systems, nor to individuals who play no meaningful role in the creative process. While national laws may vary—for example, some permit fully “computer-generated works”, assigning authorship to the person who arranged for its creation—these provisions must still comply with the overarching EU requirement that protected works reflect the author’s own intellectual creation. These national provisions therefore seem to contradict established EU law.
Furthermore, the study highlights the risk that doctrines such as the presumption of authorship—intended to simplify enforcement—may be misused in the AI context. Specifically, there is a danger that legal presumptions will be exploited to claim copyright over outputs that lack sufficient human creativity. Such developments could erode the integrity of copyright law by extending protection to de facto fully AI-generated works that lack a human author, thereby weakening the normative and doctrinal link between copyright and human creativity.
The thesis concludes that copyright protection can extend to AI-assisted computer code, but only when the human user contributes original, creative expression that is reflected in the final work. It warns that overreliance on pre-sumptions of authorship in the context of AI-generated code may undermine the normative foundations of copyright law. Without safeguards, the system risks recognizing outputs as protected “works” even where genuine human authorship is missing—potentially diluting the anthropocentric rationale of EU copyright protection. (Less)
Please use this url to cite or link to this publication:
http://lup.lub.lu.se/student-papers/record/9189385
- author
- Dib, Danny LU
- supervisor
-
- Ana Nordberg LU
- organization
- course
- JURM02 20251
- year
- 2025
- type
- H3 - Professional qualifications (4 Years - )
- subject
- keywords
- AI, Generative AI, Artificial Intelligence, Copyright, Intellectual Property, Work of Authorship, Computer Code, AI Programming Tools
- language
- English
- id
- 9189385
- date added to LUP
- 2025-06-23 16:46:01
- date last changed
- 2025-06-23 16:46:01
@misc{9189385, abstract = {{This thesis examines whether computer code generated with the assistance of artificial intelligence (AI) can qualify as a “work” under EU copyright law, and if so, who may be recognized as its author. As AI programming tools such as GitHub Copilot and Tabnine become more integrated into software development, they increasingly mediate the creative process—raising important legal questions concerning originality, human intellectual effort, authorship, and enforcement. Applying a legal dogmatic method, the thesis first establishes the EU’s harmonized criteria for copyright protection, as developed through the CJEU’s case law. To qualify as a “work”, a subject matter must fall within a recognized domain, result from human intellectual effort, reflect free and creative choices, and have these choices expressed in the final output. The thesis finds that AI-assisted computer code can meet these criteria where meaningful human input is present—particularly in the conception and redaction phases of the creative process. However, where user involvement is minimal, the resulting code may fail to satisfy the originality requirement. The analysis further shows that the threshold of originality under EU law remains relatively low, provided the author retains creative discretion despite the use of technical aids. This is especially relevant for AI-generated outputs, where the human role may be indirect or fragmented. Nonetheless, originality presupposes that the human user exercises actual creative judgment; mere reliance on fully AI-generated outputs will not suffice. In considering authorship, the thesis emphasizes that EU copyright law remains anthropocentric: only human creators may be recognized as authors. The absence of a harmonized definition of authorship at the EU level creates some legal uncertainty, particularly in the context of AI-assisted creation. Nonetheless, the CJEU’s jurisprudence has consistently linked authorship to human intellectual creation. As such, the thesis finds that current EU law does not permit the attribution of authorship to AI systems, nor to individuals who play no meaningful role in the creative process. While national laws may vary—for example, some permit fully “computer-generated works”, assigning authorship to the person who arranged for its creation—these provisions must still comply with the overarching EU requirement that protected works reflect the author’s own intellectual creation. These national provisions therefore seem to contradict established EU law. Furthermore, the study highlights the risk that doctrines such as the presumption of authorship—intended to simplify enforcement—may be misused in the AI context. Specifically, there is a danger that legal presumptions will be exploited to claim copyright over outputs that lack sufficient human creativity. Such developments could erode the integrity of copyright law by extending protection to de facto fully AI-generated works that lack a human author, thereby weakening the normative and doctrinal link between copyright and human creativity. The thesis concludes that copyright protection can extend to AI-assisted computer code, but only when the human user contributes original, creative expression that is reflected in the final work. It warns that overreliance on pre-sumptions of authorship in the context of AI-generated code may undermine the normative foundations of copyright law. Without safeguards, the system risks recognizing outputs as protected “works” even where genuine human authorship is missing—potentially diluting the anthropocentric rationale of EU copyright protection.}}, author = {{Dib, Danny}}, language = {{eng}}, note = {{Student Paper}}, title = {{EU Copyright Law and the Protection of Computer Code Created with AI Assistance}}, year = {{2025}}, }