FSF Threatens Anthropic over Infringed Copyright: Share Your LLMs Freel (www.fsf.org)

The Free Software Foundation (FSF) has stepped into the copyright dispute involving Anthropic, whose large language models (LLMs) allegedly used books from the Library Genesis and Pirate Library Mirror datasets without proper authorization. While a court ruled that training on these works might be considered fair use, the legality of downloading them for that purpose remains unresolved. Rather than pursue lengthy litigation, the parties have opted to settle, prompting the FSF to reach out to copyright holders to negotiate compensation. This development carries significant implications for the AI/ML community, particularly regarding the ethos of open access and user freedom in developing LLMs. The FSF, known for advocating free software principles, is now urging AI companies like Anthropic to release their training datasets and model details to users, emphasizing transparency and ethical use of copyrighted material. By championing the distribution of complete LLMs along with their training data and source code, the FSF hopes to foster a culture of collaboration and respect for intellectual property that aligns with the ideals of free software.