Record labels claim AI generator Suno illegally ripped their songs from YouTube

Major record labels have escalated their lawsuit against Suno, alleging that the AI startup knowingly pirated songs from YouTube to train its generative AI music models. In the amended complaint filed on September 19th, the Recording Industry Association of America (RIAA) accuses Suno of unlawfully “stream ripping” tracks on YouTube — a practice that involves turning content captured from streaming platforms into downloadable files — and circumventing measures designed to prevent unauthorized copying. 

The updated lawsuit alleges that Suno “employed code to access, extract, copy, and download” copyrighted works from Universal, Sony, and Warner, and violated YouTube’s terms of service by circumventing the platform’s “rolling cipher” encryption. This circumvention of YouTube’s technological measures “has facilitated Suno’s ongoing and mass-scale infringement,” according to the complaint, and violates the anti-circumvention provisions of the Digital Millennium Copyright Act (DMCA).

Section 1201 of the DMCA states that “no person shall circumvent a technological measure that effectively controls access to a work protected under this title.” Its uses have become extraordinarily broad in recent years, creating roadblocks for everything from unlocking phones to repairing ice cream machines. This lawsuit applies it in a way that’s closer to its original stated purpose: preventing would-be pirates from breaking DRM. Section 1201 has a process for getting exceptions that allow circumvention under certain circumstances, but so far, no specific exception exists for training AI tools.

Suno hasn’t made its training datasets public, and is vague about how the data was acquired. Suno has claimed that training its models on copyrighted materials is legally protected under fair use doctrine, an argument that’s bolstered by at least one court ruling, but not a firm legal consensus. The amended complaint is now targeting this defence by pointing to research from the ICMP publishers group that suggests Suno sourced its training data illegally by circumventing YouTube’s encryption tech. That argument is especially relevant given its similarity to Anthropic’s $1.5 billion book piracy settlement, though that case is currently on hold.

The amended complaint against Suno maintains that the AI music startup fed “decades worth of the world’s most popular sound recordings” into its AI models without authorization from record companies. The RIAA is now seeking $2,500 in statutory damages for each act of circumvention, alongside up to $150,000 per work infringed.