A federal court has dismissed a lawsuit filed by three music publishers attempting to prevent Anthropic from utilizing their song lyrics in training its AI system, which is financially supported by Amazon.
On Tuesday, Judge Eumi Lee determined that Universal Music Group, Concord Music Group, and ABKCO failed to demonstrate that their material is being unlawfully exploited for Anthropic purposes without permission or compensation. Furthermore, she highlighted potential issues with implementing and overseeing a court order that would necessitate the company to reconstruct future learning models due to the vast and seemingly limitless range of works involved in the case. If the injunction had been granted, it would have affected hundreds of thousands of songs owned by the music publishers.
In their statement, the publishers expressed high assurance that they have a strong case against Anthropic as a whole. They further noted that Anthropic has already acknowledged the validity of the publishers’ copyright claim on the infringing song lyrics by agreeing to implement certain safeguards to prevent such infringements in future, thus significantly strengthening the publishers’ position in this legal dispute.
2023 saw a legal battle unfold in Tennessee’s federal court, where I found myself at the center of accusations. Anthropic, the AI company I’m associated with, allegedly lifted lyrics from musical greats like Katy Perry, the Rolling Stones, and Beyoncé to create its AI model. To give you an idea, when prompted with the lyrics to Katy Perry’s “Roar,” our AI chatbot, Claude, came up with words remarkably similar to the original piece.
The publishers applied for a broad injunction, which if granted, would prevent Anthropic from utilizing lyrics owned by the publishers in future works.
In its decision on Tuesday, the court highlighted the vast and unpredictable realm of melodies that the restraining order could potentially cover.
In his writing, Lee stated that publishers did not provide a clear or specific method for Anthropic (the affected party) to understand the boundaries or fulfill the requirements of the court order.
Anthropic argued that removing a significant portion from their training data library would be almost impracticable. This would necessitate ongoing work to revise the collection, as well as repeatedly initiating the training process for future models.
One key point often raised by publishers is that they believe Anthropic, as well as other AI companies and lyrics aggregators, are exploiting their copyrighted materials without compensating them for the licensing opportunities these interactions could generate in the future.
The court found that the publishers failed to provide evidence showing that Anthropic’s use of song lyrics has resulted in lower license fees for lyrics websites, which it emphasized does not conflict with Claude. The court added that a significant portion of the potential damage from this behavior was already addressed in an agreement reached earlier this year between both parties, which ensures existing safeguards remain in place to stop Claude from offering lyrics to songs owned by the publishers or creating new song lyrics based on their material.
Lee additionally found that the market for training licenses has expanded since the legal action was initiated, contradicting publisher’s claims that using their lyrics to train Claude would harm their bargaining power with AI firms in future discussions. Instead of seeking an injunction, she suggested that publishers could opt for financial compensation if they win the lawsuit.
In the verdict, the court declined to address the question of fair use – a significant point in AI disputes that determines if unlicensed use of copyrighted content is acceptable. However, the court’s examination of the licensing market might offer some insights into how other courts might view this issue in the future.
Anthropic expressed satisfaction that the court declined the plaintiffs’ broad and chaotic demand for temporary relief in this matter. As the trial progresses, Anthropic is eager to clarify why utilizing copyrighted content for the purpose of training large language models falls within the scope of fair use principles under copyright law.
Read More
- Ludus promo codes (April 2025)
- Cookie Run Kingdom: Shadow Milk Cookie Toppings and Beascuits guide
- ZEREBRO PREDICTION. ZEREBRO cryptocurrency
- Grimguard Tactics tier list – Ranking the main classes
- DEEP PREDICTION. DEEP cryptocurrency
- Seven Deadly Sins Idle tier list and a reroll guide
- Shai Gilgeous-Alexander and Converse Introduce the SHAI 001
- The Entire Hazbin Hotel Season 2 Leaks Explained
- Fortress Saga tier list – Ranking every hero
- Outerplane tier list and reroll guide
2025-03-26 19:54