Sony and Warner Are Suing Anthropic Over Lyrics in Its AI
Key takeaways
- Sony Music Publishing and Warner Chappell filed a federal lawsuit against Anthropic over alleged lyric reproduction by Claude
- Sony Music Publishing is the world's largest music publisher, with catalogues including Beyoncé and Bob Dylan
- The lawsuit reportedly focuses on piracy accusations, a more direct legal angle than fair use arguments used in other AI cases
- Statutory damages in US copyright law could mean enormous financial exposure given the number of individual songs involved
The music industry has found its next major AI target. Sony Music Publishing and Warner Chappell Music have filed a lawsuit against Anthropic, the maker of the Claude AI system, accusing it of what the complaint reportedly describes as a "brazen campaign" of intellectual property theft. The case centres on song lyrics, and it's shaping up to be one of the broadest and most significant AI copyright cases to date.
The core allegation is that Claude reproduces copyrighted song lyrics when users ask for them, either quoting them directly or reproducing substantial portions without authorisation. Music publishing companies, unlike record labels, own the underlying compositions rather than the recordings. If an AI can reproduce protected lyrics on demand, that undermines the licensing infrastructure that publishers have spent decades building.
What the Lawsuit Is Claiming
Sony Music Publishing and Warner Chappell are not small players in this space. Sony Music Publishing is the largest music publisher in the world by some measures, representing catalogues that include artists from Beyoncé to Bob Dylan. Warner Chappell is similarly vast. Together, they control an enormous portion of the song lyrics that have ever been written down and commercially released.
The lawsuit, filed in US federal court, allegedly homes in on piracy as a specific accusation rather than the more ambiguous fair use arguments that have characterised some earlier AI copyright disputes. The companies are claiming that Anthropic's training process ingested copyrighted lyric data without permission, and that the resulting model reproduces that data in a way that constitutes direct infringement.
This is meaningfully different from some earlier AI lawsuits. The visual art and news content cases often hinge on whether training on copyrighted material constitutes infringement at all, which is genuinely uncertain law. The lyrics angle is more direct: if a user asks Claude to write out the lyrics to a specific song and Claude does it, that output is potentially a straightforward reproduction of protected text.
Anthropic's Position and the Industry Context
Anthropic has not yet issued a detailed public response to this specific lawsuit. The company has generally positioned itself as a safety-focused AI developer and has engaged with some content creator concerns around its training data. But the scale of what Sony and Warner are alleging makes a quiet settlement complicated.
The timing is also notable. This comes as the broader AI industry is under increasing legal scrutiny from rights holders across music, publishing, and visual media. The New York Times has an ongoing case against OpenAI. Getty Images has sued Stability AI. Universal Music Group has been vocal about wanting AI training licences to become standard practice. The music publishing industry has been watching these cases develop and now appears to be moving aggressively on its own.
For Anthropic, this is a serious distraction and a potential financial liability. Copyright damages in the US can be significant, particularly when statutory damages are available, which they often are for registered works. Given that Sony and Warner between them represent hundreds of thousands of individual songs, each potentially constituting a separate infringement, the theoretical damages exposure is enormous.
What It Means for AI Development
This case is part of a larger reckoning that the AI industry has been aware was coming for several years. The training data that made large language models capable of writing naturally about music, quoting artists, and engaging with popular culture was largely scraped from the internet without negotiated permissions. The question of whether that was legal is still being litigated in multiple jurisdictions.
For users of AI tools, the practical implication might be that AI assistants become more cautious about reproducing lyrics specifically. Some already are. OpenAI's ChatGPT, for instance, has been inconsistent about lyrics reproduction for some time, likely due to internal content policies developed in anticipation of exactly this kind of legal pressure.
The more interesting long-term question is whether this results in licensing deals between AI companies and music publishers, similar to the deals that streaming platforms negotiated with labels in the early days of Spotify. That outcome would benefit publishers financially and might actually make AI music tools more capable rather than less. But getting there requires the lawsuit to first establish that the liability is real, which is exactly what Sony and Warner appear to be trying to do.