AI training and Copyright

26 March 2025

Judge denies a preliminary injunction sought by music companies claiming that Anthropic was harming them by using copyrighted material to train its AI chatbot, Claude

Anthropic, an AI safety and research company, recently achieved a legal victory when a U.S. federal judge denied a preliminary injunction sought by Universal Music Group (UMG) and other music publishers. The publishers had alleged that Anthropic's AI chatbot, Claude, reproduced copyrighted song lyrics from at least 500 songs without authorisation, constituting copyright infringement. They sought to prevent Anthropic from using their lyrics to train future AI models. However, Judge Eumi K. Lee ruled that the publishers failed to demonstrate "irreparable harm" from Anthropic's use of the lyrics, stating that any potential harm could be addressed through monetary damages rather than an injunction.

This ruling is part of a broader legal discourse on AI and copyright. Previously, in January 2025, Anthropic agreed to implement "guardrails" to prevent Claude from generating copyrighted lyrics, partially resolving aspects of the lawsuit. Despite this, the publishers remain confident in their broader case against Anthropic, emphasising the importance of protecting creators' works. ​

The case underscores the ongoing debate over the use of copyrighted materials in training AI models and the legal definitions of "fair use." Similar lawsuits are unfolding, with publishers and AI companies navigating the complex intersection of technology and intellectual property rights. (article)