Suno snatched millions of songs from YouTube, Genius, and Deezer

Leaked data from a hack has confirmed that the AI music platform Suno trained its models by scraping millions of audio files and lyrics from sources including YouTube Music, Deezer, and Genius. While Suno has previously maintained that its models are trained on publicly available internet content under fair use, these findings support allegations made in RIAA lawsuits that the company utilized unauthorized methods, such as stream ripping, to acquire copyrighted material. The leaked materials include scraping instructions and internal logs detailing the volume of consumed media. Additionally, the breach exposed customer data, though Suno contends that the incident primarily involved outdated code and that no sensitive financial information was compromised.

(Source：The Verge)