Elsevier vs. Meta: first science publisher sues over scraped research papers (www.nature.com)

Elsevier, along with major publishers Hachette and Macmillan, has initiated a class-action lawsuit against Meta, claiming that the tech giant unlawfully used copyrighted materials to train its large language model, Llama. This marks a significant legal development as it is the first lawsuit of its kind launched by major publishing houses, emphasizing their stance against perceived violations of intellectual property rights. The lawsuit alleges that Meta accessed copyrighted works, including scientific papers, through sources like the Common Crawl data set and file-sharing platforms such as LibGen and Sci-Hub. This legal action highlights ongoing tensions between AI companies and content creators regarding copyright issues, against a backdrop of similar lawsuits filed by authors and media organizations. The outcome of this case could set a crucial precedent for the AI/ML community, specifically addressing the legality of using copyrighted materials for training models. Meta plans to defend itself by arguing that such usage could qualify as "fair use" under U.S. copyright law. As the field of AI continues to evolve, the implications of this lawsuit could reshape how companies approach data sourcing and intellectual property in the development of AI technologies.