Sony Music Entertainment escalates its legal war against AI music generator Udio, filing a fresh lawsuit alleging the startup trained its models on over 30,000 illegally copied songs. The complaint names major artists including Beyoncé, Harry Styles, and Elvis Presley whose work allegedly fed Udio's machine learning system without permission or compensation.
This marks Sony's second major offensive against Udio. The entertainment giant, joined by Universal Music Group and Warner Bros. Records, first sued the AI startup in June 2024, establishing a pattern of aggressive legal action against generative music platforms. Sony targets Udio specifically for allegedly circumventing content protection measures to scrape its catalog at scale.
The lawsuit positions Sony among the music industry's most vocal opponents of unauthorized AI training. The company has consistently argued that generative music tools built on copyrighted material without licensing agreements violate both copyright law and artist rights. This fresh complaint suggests Sony views previous legal filings as insufficient deterrents.
Udio operates in a contested space where AI music generators promise quick, low-cost composition tools for creators and businesses. The startup competes directly with Suno, another AI music platform facing identical allegations from the same rights holders. Both companies have maintained that their training processes fall within fair use protections, though courts have yet to definitively rule on this argument in the music AI context.
The broader legal landscape remains unsettled. Generative AI companies argue their technology trains on publicly available data for transformative purposes. Rights holders counter that commercial AI tools built on unlicensed copyrighted content constitute straightforward theft. Sony's sustained litigation strategy suggests the label believes courtroom victories offer the fastest path to forcing licensing negotiations.
The stakes extend beyond Sony's catalog. This litigation shapes whether AI music generators can operate viably under current copyright frameworks or must pivot toward licensed training data. Industry observers watch these cases