Sony Music Publishing and Warner Chappell Music filed a federal lawsuit on August 28 against Anthropic, CEO Dario Amodei, and co‑founder Benjamin Mann. The complaint alleges that Benjamin Mann personally downloaded five million books via torrent from LibGen in 2021 and a further two million in 2022, purportedly at Amodei’s direction.
Claims in the complaint
The plaintiffs cite internal documents that described LibGen as “sketchy AF” and included the note “We don't want it to be known.” Rather than focusing primarily on how any collected material was used in model training, the suit targets the source of the material and who authorized its acquisition.
Sony Music Publishing and Warner Chappell Music seek statutory damages that could apply to tens of thousands of works; the complaint notes potential damages of up to $150,000 per work.
Why this matters
The lawsuit represents a shift in the ongoing dispute over AI and copyright. For years the debate centered on whether training on copyrighted material constitutes infringement or can be fair use. The current legal approach from these plaintiffs asks a different question: where did the data come from, and who approved taking it?
The filing references a distinction made by the court in Bartz: training on legally obtained works may be treated differently under fair use analysis, while obtaining works through piracy can itself constitute infringement. By concentrating on acquisition and on individual decision‑makers, the plaintiffs move beyond questions about downstream model use.
Potential consequences
By naming company founders personally, the plaintiffs signal a willingness to pierce the corporate shield. If a court allows personal liability for data‑acquisition decisions, founders and executives who authorized or approved data‑collection pipelines could face direct legal exposure. That would broaden the targets of copyright litigation beyond companies to the individuals who managed data procurement.
Summary
The August 28 lawsuit by Sony Music Publishing and Warner Chappell Music accuses Anthropic founders of directing or carrying out mass downloads from LibGen—allegedly five million books in 2021 and two million in 2022—and relies on internal documents that acknowledge the questionable source. The case reframes the copyright conflict from a focus on how data were used to a focus on how data were obtained and who approved the acquisition, with possible implications for piercing corporate protections and increasing personal liability for AI founders.
