Speaking with entertainment trade publication The Ankler, litigation partner Mark Lezama offered his perspective on the evolving issues of artificial intelligence and copyright infringement in the film and television industry.
The article, “TV Writers Found 139,000 of Their Scripts Trained AI. Hell Broke Loose,” focused on the recent discovery of a dataset used to train artificial intelligence Large Language Models (LLMs) that extracted data from thousands of film and television scripts.
Addressing a central question at play – whether this form of training constitutes copyright infringement versus fair use – Lezama commented that the courts have yet to weigh in. However, he shed light on the arguments on both sides of the issue, including what defines a copyrighted work and how market value of original works could impact fair use analysis.
Lezama explained that weighing fair use could include understanding how use of a copyrighted work impacts the original work’s market value and that proponents of the current AI training model might argue there is no interference. He noted that supporters could claim using data from these scripts does not directly compete with the original work, as the goal is to copy the data for training purposes, not to create a script that will compete for production.
Read Lezama’s full comments here [subscription required].
