Authors’ Rights in the Context of Selected Models for AI Training Cover Image

Trenowanie wybranych modeli sztucznej inteligencji a uprawnienia twórców
Authors’ Rights in the Context of Selected Models for AI Training

Author(s): Iga Bałos
Subject(s): Law, Constitution, Jurisprudence, Civil Law, Social Informatics
Published by: Oficyna Wydawnicza KA AFM
Keywords: AI generated works; copyright; AI; text and data mining;

Summary/Abstract: Artificial Intelligence (AI) is no longer about automatization of routine or repetitive tasks. Its application extends far beyond plain computational power and efficiency. “AI-generated works” is one of the most popular research areas in this filed. It raises the question about intelligent systems singularity and their true capability to perform human-like tasks. This article examines selected models for AI training which generate works mimicking the unique style of given artists provided their works serve as input data. Certain authors, perhaps especially writers, are reluctant to accept such forms of the exploitation of their works. Even though classification of AI-generated works is a challenge, significant legal issues arise at the earlier stage. Does text and data mining (TDM) infringe the copyright monopoly? Is using a collection of works of a single author as training data affects the lawfulness of a process? Are authors entitled, on the basis of their economic or moral rights, to oppose such practices?

  • Issue Year: 25/2019
  • Issue No: 2
  • Page Range: 15-28
  • Page Count: 14
  • Language: Polish