Source: Waxy
Suno's reliance on scraped copyrighted material—from YouTube, Deezer, and Genius—creates legal exposure that extends beyond the fair use debate. Even if courts permit AI training on copyrighted works, the method of acquisition itself (unauthorized scraping rather than licensed access) represents a distinct liability. The distinction separates the question "can you train on copyrighted data?" from "can you steal that data to train on?"—and the latter has clearer legal precedent against tech companies. Future AI companies will likely need to negotiate licenses or use synthetic and licensed datasets upfront, rather than scrape now and litigate later.