AI Company Anthropic Sued by Universal Music, Concord, and ABKCO—Here’s Why

AI company Anthropic has been sued by music publishers Universal Music, ABKCO, and Concord Publishing over alleged copyright infringement of their songs.

According to Reuters, this lawsuit filed in Tennessee federal court on Wednesday appeared to be the first against Anthropic and its first case over song lyrics.

The publishers accused Anthropic of using an "innumerable" amount of copyrighted song lyrics to train its chatbot Claude and posting the lyrics in its prompted answers without permission. The lawsuit claimed this was a "clear violation against well-established copyright laws."

By unlawfully copying and disseminating vast amounts of copyrighted works, the AI startup funded by tech giants Google and Amazon allegedly violated these laws on a "systematic and widespread basis."

The publishers claimed that Anthropic used lyrics from more than 500 songs to train its chatbot, which includes "God Only Knows" by the Beach Boys, "Gimme Shelter" by the Rolling Stones, "Halo" by Beyonce, and "Uptown Funk" by Mark Ronson and Bruno Mars.

The publishers are reportedly seeking potentially tens of millions of dollars in damages from Anthropic and an order to stop the alleged infringement.

Anthropic's 'Unlawful Exploitation'

Anthropic's "unlawful exploitation" was explained in the lawsuit through various instances. Universal Music and its co-plaintiffs noted that the AI company keeps the origins of the content used to train its AI a secret.

According to the lawsuit, Anthropic only provided limited information, saying that its "Claude models are trained on a proprietary mix of publicly available information from the Internet, datasets that [they] license from third party businesses, and data that our users affirmatively share or that crowd workers provide," and that the text on which Claude 2 was trained continued through early 2023 and is 90% English.

Because Anthropic was allegedly aware that it was duplicating copyrighted information without permission from the copyright owners, the company reportedly refused to be transparent, not sharing the materials it used to teach Claude.

Another instance claimed in the lawsuit was how the company's AI model responds with evidently exact or nearly exact copies of the publishers' copyrighted lyrics when prompted by users.

Anthropic also allegedly failed to indicate that it will take any actions to remove copyrighted content, even though it "cleans" the text it ingests to eliminate offensive language and other elements it wants to exclude from its training corpus.

Anthropic's AI model has also been claimed to be capable of creating lyrics for new songs using lyrics from existing copyrighted songs. In some instances, bits of a copyrighted work may be combined with portions of another copyrighted piece, thus "inconsistent" and damaging to its intended form.

Universal Music and Co-Plaintiffs on AI

It was stated in the lawsuit that the music publishers viewed AI positively. They clarified that the companies "embrace innovation and recognize the great promise of AI when used ethically and responsibly."

However, they said that "Anthropic violates these principles" as it failed to "abide by well-established copyright laws, just as countless other technology companies regularly do."

"As a result of Anthropic's mass copying and ingestion of Publishers' song lyrics, Anthropic's AI models generate identical or nearly identical copies of those lyrics, in clear violation of Publishers' copyrights," the music publishers noted.