A lyrics lawsuit against Anthropic is still open after two years

The system

Anthropic's Claude models are large language systems, not music-audio generators, but the company still faces a music-industry copyright suit because its models were trained on text scraped from the internet, and song lyrics are text. Concord Music Group, joined by Universal Music Corp., ABKCO Music, Capitol CMG and other publishers, filed a complaint against Anthropic PBC in the U.S. District Court for the Middle District of Tennessee, Nashville Division, on 18 October 2023. The publishers allege that Anthropic copied and disseminated the lyrics to compositions they own or control while building Claude, without a license. This is a distinct case from any lawsuit over music-audio generators such as Suno or Udio.

What the documents establish

The complaint states the claim in plain terms: Anthropic 'unlawfully copies and disseminates vast amounts of copyrighted works — including the lyrics to myriad musical compositions' to train and operate Claude, which the publishers' filing characterizes as infringement rather than innovation. That Tennessee docket shows a termination date of 24 June 2024, but the dispute did not end there. A case with the same lead plaintiff and defendant, Concord Music Group, Inc. v. Anthropic PBC, was filed in the Northern District of California on 28 January 2026, now naming additional publishers including Warner Chappell Music and Sony Music Publishing, and individual Anthropic co-founders. Its docket, current as of retrieval, records a stipulated dismissal of one claim, vicarious infringement, in May 2026, a stay tied to related actions in April 2026, and continuing filings into September 2026: the case has not been resolved on the merits and no ruling on liability has issued.

Craft and rights

For a songwriter or publisher, this case tests whether the same infringement logic that applies to song audio also reaches lyrics ingested as plain text, a question with consequences for every AI company that trains language models on scraped web text containing copyrighted lyric transcriptions. What remains a corporate decision is licensing: the publishers' own complaint frames the dispute as Anthropic's failure to seek permission, not a claim that lyrics can never lawfully be used to train AI. This is an editorial observation: the case's expansion to include more publishers and individual defendants suggests the parties see the underlying legal question as unsettled enough to keep litigating rather than settle quickly, unlike Universal's separate, later licensing deals with AI music platforms.

Outcomes and open questions

No court has ruled on whether Anthropic's use of lyrics is infringing or fair use. The case's procedural history, moving districts and adding parties, means any claim about its outcome must be checked against the current docket rather than assumed from the original 2023 filing.

• Has this docket produced any ruling on the merits since this report's retrieval date?
• Does your catalogue include lyrics that could plausibly appear in a large text-training corpus?
• What would a ruling either way mean for licensing text, as distinct from licensing audio, to AI developers?

Litigation over lyrics and litigation over generated audio raise related but separate legal questions, and conflating them risks overstating what any single ruling, when one arrives, will actually settle.