Anthropic agrees to stop Claude Chatbot’s ability to spit Lyrics From Copyright Songs

Anthropic has arrived at a settlement and agreed to stop showing users music lyrics based on copyrighted songs from multiple music publishers. Back in 2023, the AI ​​company was sued from Universal Music Group, Concord Music Group, and others after it was found that its chatbot Claude would return lyrics to songs like Beyoncé’s “Halo” when asked.

The entertainment industry is one of the most litigious and fights vigorously to defend its copyrights – just look at historical cases, from the destruction of Napster to the multi-year legal battle that Viacom fought against to YouTube. More recently, the popular lyric annotation website Rap Genius (now called Genius) was sued by the National Association of Music Publishers for the reproduction of copyrighted song lyrics.

The music publishers suing Anthropic acknowledge that other websites such as the music annotation platform Genius distribute lyrics online, but they noted that Genius eventually began paying a license to publish them on its website.

In this latest suit, the music publishers claimed that Anthropic scraped lyrics from the web and intentionally removed watermarks that are placed on lyric websites to help identify where the copyrighted material was published. After Genius started licensing lyrics from music publishers, it’s smart extra apostrophes inserted in the lyrics so that, in the event that the material has been copied inappropriately, Genius will know that the material that he has explicitly paid for has been stolen and will be able to request its removal.

Anthropic did not concede the claims, but as part of the settlement agreed to better maintain guardrails that prevent its AI models from infringing copyrighted material. It also works in good faith with music publishers when it is found that the guardrails are not working.

Anthropogenic defended the act of using song lyrics and other copyrighted materials for the training of AI models, saying The Hollywood Reporter“Our decision to enter into this stipulation is consistent with these priorities. We continue to expect to demonstrate that, consistent with existing copyright law, the use of potentially copyrighted material in the training of generative AI models is a quintessentially fair use. This argument has been central to the AI ​​companies’ defense of the copyrighted material that appears in their models. Defenders argue that remixing copyrighted content from websites like the New York Times constitutes fair use, as long as it has been materially altered through derivative works.

News and music publishers disagree, and the lawsuit against Anthropic is far from over. The music publishers are still seeking a court injunction preventing Anthropic from forming future models on any copyrighted music lyrics.

The concern of abuse comes from the potential of Anthropic’s models to be used in the generation of music that causes a musician to lose control of their art. This is not an unfounded concern, as it has been widely speculated that OpenAI imitated Scarlett Johansson’s voice after she refused to provide her voice for its AI voice model.

Tech companies like OpenAI and Google make their money on platforms and network effects, not on selling copyrighted material, which has always led to this tension between Hollywood and Silicon Valley. Art is just “content” meant to serve the larger purpose of generating revenue and selling ads. The AI ​​slop that fills Facebook today is representative of how tech companies see everything as interchangeable.

Editors like the Times have fought high-profile battles against the likes of OpenAI in court to prevent them from sucking up copyrighted material. OpenAI has tried to respond by licensing material from some companies, and another AI player, Perplexity, has begun testing a revenue-sharing model. But publishers want more control and not be forced into these shaky deals that could end at any moment and also drive people away from their websites. Which is all to say, this is far from the end of the story when it comes to disputes over copyrighted material in major language models.