In the midst of the burgeoning AI-related legal landscape, the New York Times has initiated a copyright infringement lawsuit against Microsoft and OpenAI. The central contention revolves around the alleged copyright infringement when training large language models (LLMs) on material copyrighted by the Times.
OpenAI has staunchly defended its position, asserting that the practice of training AI models using publicly available internet materials falls within the bounds of fair use, citing support from established precedents. Notably, the Library Copyright Alliance (LCA) has backed this stance, emphasizing that the ingestion of copyrighted works for AI training databases aligns with fair use principles.
In response to efforts seeking to amend copyright law for generative AI, LCA, as advocates for fair use, free speech, and freedom of information, drafted principles to maintain the balance of copyright law. The LCA principles underscore the flexibility and robustness of current copyright law in addressing AI-related issues without necessitating amendments. The distinction between input to train an LLM and potentially infringing output is a critical aspect outlined in these principles.
Scholars and librarians are actively invested in preserving the transformative fair use treatment of training AI models. Restricting researchers to training AI on public domain works would limit the scope of inquiries, hindering non-generative, nonprofit educational research methodologies like text and data mining. The implications extend to a diminished ability to explore modern in-copyright materials, impacting the relevance of research to contemporary concerns.
Despite ongoing legal battles, the precedent established by cases such as Authors Guild v. HathiTrust and Authors Guild v. Google underscores the transformative fair use of AI. The lawsuits raise critical questions at the intersection of copyright law and AI, with ongoing monitoring by the LCA. The evolving landscape, influenced by the pending cases and the upcoming US Copyright Office Study, will shape the future interpretation of copyright law in the realm of AI.
More: https://www.arl.org/blog/training-generative-ai-models-on-copyrighted-works-is-fair-use/
