Anthropic Claims Fair Use Defense in Music Publishers' Copyright Battle Over AI Training Data
Anthropic has formally responded to music publishers' preliminary injunction request in their ongoing copyright infringement dispute, asserting that using copyrighted content to train large language models (LLMs) constitutes fair use.
Microchip with AI text overlay
The dispute began when Universal Music, Concord, and ABKCO sued Anthropic, claiming unauthorized use of their compositions in AI training and alleging that Anthropic's Claude AI assistant reproduced song lyrics without attribution.
Key points from Anthropic's 40-page response:
- Training LLMs on copyrighted content qualifies as fair use
- Publishers haven't demonstrated irreparable harm
- Similar cases nationwide proceed without preliminary injunctions
- Anthropic offered to implement additional safeguards
- Song lyrics aren't typical Claude AI outputs
Anthropic argues that requiring licenses for AI training would make general-purpose AI tools impossible, stating: "One could not enter licensing transactions with enough rights owners to cover the billions of texts necessary to yield the trillions of tokens that general-purpose LLMs require."
Additional arguments include:
- Venue challenge (citing no relevant connection to Tennessee)
- Claims that publishers themselves generated the alleged infringing copies through their prompts
- Publishers delayed several months before taking legal action
Anthropic logo on black background
Anthropic maintains that if the publishers ultimately prevail, monetary damages would be sufficient remedy, eliminating the need for preliminary injunctive relief.
Related Articles
Diplo Revenge Porn Case Dismissed After Judge Orders Accuser Identity Reveal
