Suno 抓取了 200 萬首 YouTube 歌曲——並付費給 Bright Data 來執行
被洩露的資料集顯示 Suno 從 YouTube Music 和 Genius 抓取了超過 200 萬個片段及歌詞來訓練其 AI 模型，證實了 RIAA 的指控。
hacked source code reveals suno scraped 2,013,545 music clips from youtube, alongside thousands of hours from Deezer, Genius, and Pond5, to train its generation models. The documents detail scraping operations from 420,000 podcasts and the use of Bright Data proxies to circumvent protections. This leak validates the Recording Industry Association of America's claims of massive copyright infringement, proving suno accessed restricted content despite claiming fair use. The breach also compromised data for hundreds of thousands of users, exposing specific payment and personal info.