Major record labels allege Suno used “stream-ripping” to scrape copyrighted songs from YouTube, escalating their ongoing AI music lawsuit.
Major record labels Universal Music Group, Warner Music Group and Sony Music Entertainment have moved to expand their lawsuit against AI music company Suno by adding a piracy claim, alleging the startup illegally downloaded copyrighted songs from YouTube to train its generative models.
In proposed amended court filings dated Sept. 19, the labels assert they recently confirmed Suno engaged in “stream-ripping,” a technique that circumvents YouTube’s technical safeguards and, the labels argue, violates the Digital Millennium Copyright Act (DMCA).
If proven, the labels seek statutory damages of up to $2,500 for each alleged act of piracy.
The proposed amendment cites an exclusive Billboard report that exposed private datasets showing how both Suno and rival Udio allegedly scraped music from the internet at scale.
According to the labels’ filing, these datasets demonstrate the mass collection and storage of copyrighted works, and the stream-ripping method specifically undermined YouTube’s encryption mechanisms intended to prevent downloads.
The Recording Industry Association of America (RIAA), speaking for the labels, accused Suno of deploying a “game of deception” by obscuring whether copyrighted recordings were used to train its models and how those works were obtained.
Suno has not yet publicly responded to requests for comment about the new piracy allegation. Notably, the labels have not moved to add the same piracy claim against Udio, another prominent AI music company named in the broader litigation alleging unauthorized use of recorded music for model training.
The labels’ shift to emphasize piracy claims follows a high-profile development in related AI copyright litigation: Anthropic’s announced settlement of $1.5 billion with authors who accused the company of building its training corpus from pirated books.
That deal has encouraged other plaintiffs to bolster their suits with piracy-related theories, arguing that mass scraping and unauthorized downloading create an additional basis for damages beyond more traditional copyright infringement claims.
At the same time, legal questions about whether AI training itself constitutes fair use remain central and hotly contested. In June, U.S. District Judge William Alsup — overseeing the authors’ case against Anthropic — ruled that storing pirated books could subject a company to liability while also indicating that the act of training an AI on copyrighted text might be analyzed under fair use doctrines.
That nuanced stance leaves open the possibility that copying for model training could, in some circumstances, be considered transformative and thus permissible, even while certain methods of acquisition (like circumventing technological protections) could be separately unlawful.
The labels are expected to press both infringement and DMCA-based piracy claims through discovery and subsequent litigation phases.
Suno and Udio have maintained that training on existing songs is transformative; the courts will ultimately decide whether those defenses succeed and whether piracy-based damages will play a major role in shaping liability and remedies in the rapidly evolving intersection of AI and copyright law.