Anthropic comes to agreement with music publishers concerning lyric issue

Anthropic has agreed to the stipulation that a few components of a copyright-infringement case filed against the creator of the Claude AI model have been settled; the accused shared protected song lyrics. It was certified on Thursday by a United State’s District Judge Eumi Lee whereby, Anthropic would employ existing guardrails in training of future AI models together with developing a process by which music publishers can put restraints on the AI assistance where they suspect infringement of copyright. Music companies such as Universal Music Group, ABKCO, Concord Music Group, as well as song publisher Greg Nelson music sued Anthropic in a copyright infringement lawsuit in October 2023 accusing the organisation of training its AI on lyrics from over 500 songs in its database.

In its complaint, when Claude was “asked to sing along to songs such as Beyoncé’s ‘Halo,’ Mark Ronson’s ‘Uptown Funk,’ and ‘Moves like Jagger’ by Maroon 5 [it] produced responses that contain all or significant portions of those lyrics.” Acknowledging to the fact that many websites like these are already distributing lyrics online, the music publishers pointed out the fact that such sites use the protected works only after they paying a licensing fee. Anthropic intentionally deprived or modified copyright management information of the affected songs when they took data from those sites to train their AI models, the lawsuit filing stated.

In the deal inked on Thursday, Anthropic assures the partners it will preserve the safeguards it put in place to ensure that the AI system does not violate copyright law. Anthropic will also use the guardrails developed for this work on any other AI systems that it may build in the future. To address the situations, in which the safeguards will be deemed ineffectual, both music publishers and Anthropic will negotiate them in good faith thereafter bringing the issues to the court for resolution.

Anthropic was asked to comment on the ‘DKenmari’ bot, which uses Claude to generate comments and lubricate online discourse; Claude “isn’t designed to be used for copyright infringement and we have numerous processes in place to prevent such infringement,” the release stated. That is why our decision to enter into this stipulation is also a noble cause and consistent with those priorities. We still eagerly await our chance to demonstrate that training generative AI with potentially copyrighted content is an allowable fair use under current copyright law .

The music publishers who initiated the complaint also ask for a preliminary injunction to prevent Anthropic from training future models on the protected lyrics of the songs, and the court’s decision is supposed to be announced in several months.

/ This way, music publishers will engage Anthropic to ensure that the bars set by AI ensure that lyrics cannot be violated.