Anthropic AI Copyright Case Involves ‘Guardrails’ for Song Lyrics
A federal judge has approved a tentative agreement in a lawsuit accusing artificial intelligence company Anthropic of using copyrighted lyrics without authorization or payment when training its system.
Music publishers including Universal Music Group, Concord Music Group and ABKCO submitted a Copyright infringement litigation A lawsuit was filed against Anthropic in 2023. The lawsuit alleges that Anthropic’s Claude Eyre The chatbot was trained on at least 500 song lyrics by major artists, including Beyoncé and Katy Perry, and generated query results, without Anthropic receiving permission or compensating copyright holders.
On Thursday, U.S. District Judge Eumi Lee signed an agreement The agreement between the publisher and Anthropic requires the AI company to maintain its current copyright guardrails throughout the proceedings, which means no lyrics from old or new songs will be provided.
The agreement says there’s nothing to prevent Anthropic from “expanding, improving, optimizing or changing the implementation of such guardrails, provided that such changes do not materially impair the efficacy of the guardrails.” It also allows a publisher to publish if it deems the guardrails insufficient. Business owners keep Anthropic informed at any time, at which time the company must address their concerns.
The court is expected to decide in the coming months whether songs owned by music publishers can be used to train artificial intelligence systems.
The case highlights the tensions between developers of generative AI and copyright holders, and not just in the music industry. Publishers including the New York Times sued ChatGPT maker OpenAI and Microsoft, and the Wall Street Journal and Dow Jones sued Perplexity, all of which Alleged copyright infringement. Meanwhile, other publishers have struck licensing deals with artificial intelligence companies such as OpenAI and Meta.
Anthropic is backed by Amazon, Google, Salesforce and others, and was founded in 2021 by former OpenAI researchers. The data sets accepted by its system may include copyrighted material such as song lyrics, raising questions about fair use and appropriate licensing.
In a statement shared with CNET, the company emphasized its commitment to copyright compliance, including entering into the current agreement.
“Claude was not designed to infringe copyright, and we have a number of processes in place to prevent such infringement,” a spokesperson said. “We continue to look forward to demonstrating that, under existing copyright law, the training of generative AI models Use of potentially copyrighted material is typically fair use.”
Universal Music Group, Concord Music Group and ABKCO did not immediately respond to requests for comment, but the original lawsuit accuses Anthropic of “unlawfully copying and distributing a substantial number of copyrighted works.”
“Like developers of other technologies before them, from printing presses to copiers to web crawlers, AI companies must comply with the law,” the lawsuit states.
Anthropic continues to face other legal challenges. Earlier this year, the company was named in a copyright infringement class action lawsuit accusing Anthropic of “building a multibillion-dollar business by stealing hundreds of thousands of copyrighted books.”
