California District Court The solution is in the proceedings against Anthropic, which was accused of violation of copyrights when using legally purchased books and illegal collections of books from pirate sites for teaching a large language model Claude. The court ruled that the use of legally purchased books for teaching the model corresponds to the principle of conscientious use and obtaining permission from the authors in this case is not required, since books are used to analyze and generate a new text, and not for direct copying of the originals.
At the same time, the court recognized the copyright to use books loaded from pirate sites that cannot be replaced by public libraries. The amount of compensation for the use of pirate content for teaching the Claude model will be determined in a separate trial.
The court also rejected the arguments of the plaintiffs that large language models undermine competition among the authors, indicating that the copyright law was designed to protect original copyright work, and protect the authors from competition. According to the judge, complaints about increasing competition due to teaching language models can be compared with complaints that teaching schoolchildren will lead to an increase in competing work.