A judge has signed off on an agreement that partly resolves the lawsuit between AI firm Anthropic and Universal Music Group, Concord Music Group, and a slew of other record labels and subsidiaries, asserting Anthropic can’t recite the publishers’ copyrighted song lyrics or use them to make new ones.
The legal stipulation The Hollywood Reporter shared on Thursday states that Anthropic will continue to block its AI products (like its Claude chatbots) from reproducing protected song lyrics and will apply these rules to any future products, as well. If the record labels aren’t happy with the AI’s responses or outputs, they can ask Anthropic to make additional changes to prevent “derivative works” or reproductions of lyrics.
At time of writing, Claude refuses requests to recite lyrics from Universal Music artists like Taylor Swift as well as copyrighted text from well-known book authors outside of the lawsuit, suggesting Anthropic has also already applied these copyright-related restrictions to more categories beyond just music. Based on the model’s responses in a quick test, however, it appears that Claude still uses the copyrighted material internally to offer its own analysis of the songs instead of reproducing them verbatim for a human user.
Anthropic said in a statement that Claude was not “designed” for copyright infringement purposes. “We have numerous processes in place designed to prevent such infringement,” it said, adding: “Our decision to enter into this stipulation is consistent with those priorities. We continue to look forward to showing that, consistent with existing copyright law, using potentially copyrighted material in the training of generative AI models is a quintessential fair use.”
Claude refuses a request to share song lyrics. (Credit: Kate Irwin/PCMag via Claude.ai )
The aforementioned music publishers sued Anthropic in 2023, alleging widespread copyright infringement. “A defendant cannot reproduce, distribute, and display someone else’s copyrighted works to build its own business unless it secures permission from the rightsholder,” the labels’ attorneys wrote in the original complaint.
But the music industry isn’t the only one concerned about how AI models may be using their copyrighted content without permission or compensation. Last year, three authors sued Anthropic, alleging copyright infringement because they found their books in a large dataset known as “The Pile” on which AI models including the Claude models were reportedly trained. News outlets have also been fighting other AI firms over copyright concerns, like The New York Times‘ legal battle against Microsoft and OpenAI or Condé Nast’s letters to Perplexity AI.
Recommended by Our Editors
Last month, Anthropic made its Claude 3.5 Haiku available via the Claude.ai website and its mobile app. The AI firm has previously claimed that its 3.5 Sonnet model is more powerful than OpenAI’s GPT-4o.
PCMag has reached out to Anthropic for comment.
Get Our Best Stories!
This newsletter may contain advertising, deals, or affiliate links.
By clicking the button, you confirm you are 16+ and agree to our
Terms of Use and
Privacy Policy.
You may unsubscribe from the newsletters at any time.
About Kate Irwin
Reporter
