Bibliographie sélective OHADA

Generative AI has heralded a new age of possibilities and significantly altered the landscape of creation. What was once a niche tool has quickly become a de facto tool that everyday users employ to express themselves, particularly due to the rapid integration of such models into popular digital platforms. Accompanying the extensive usage of Generative AI are concerns from rightsholders regarding the tool’s ability to facilitate rapid and extensive copyright infringement. While complete regurgitation has clear copyright infringement concerns, the precise legality remains unclear for AI-generated output that has the potential to benefit from copyright exceptions and limitations. In efforts to reduce its liability, Generative AI models have introduced safety ‘guardrails’ to prevent, restrict, or alter prompts that it deems would generate output that might infringe copyright. While guardrails serve an important purpose in the face of blatant infringement, these also block or revise prompts which may benefit from exceptions and limitations, and therefore do not infringe copyright. Given these developments, the question of whether AI-generated output can benefit from exceptions and limitations to copyright becomes pertinent. Particularly in the realm of parody and pastiche, restrictions on such output pose serious freedom of expression concerns. This paper first explains how the applicability of these exceptions to GenAI output is independent from the questions of legality in the input stage, and ought to be kept that way. Thereafter, it examines the question of whether these exceptions of parody and pastiche under the EU’s copyright regime can apply to AI-generated outputs, and to what extent they apply.