Artists Discover Their Music in AI Training Database

A searchable database published by The Atlantic through its AI Watchdog project now allows artists to check whether their work was used to train AI music generators like Suno and Udio. Within hours of the tool going live, producers across the electronic music community began searching their own names and discovering their catalogs had been swept into training models without consent.

Over 21 Million Recordings Mapped

The database draws from four datasets and maps more than 21 million copyrighted recordings currently circulating among AI developers. Many were distributed through links that reportedly let developers pull music while bypassing paywalls and mechanisms that normally compensate artists. Training data has long been a closely guarded secret, but the tool marks the first time musicians can see their own songs inside the machine.

Dance Producers Report Triple-Digit Tallies

A range of influential dance music producers, including SG Lewis, tyDi, and Varien, posted tallies showing that more than 100 of their own songs had been pulled into the datasets from their respective catalogs. Prolific hip-hop producer Kenny Beats, who previously produced electronic dance music and toured as LOUDPVCK, specifically set his sights on Suno in a post on X, calling the platform's proprietors "true losers."

Coalition Demands Transparency and Consent

On Monday, a broad coalition of artists, songwriters, and managers published an open letter pressing labels and publishers over their AI licensing deals. The letter demands no default opt-ins, no forced AI clauses, and no use of an artist's work or likeness without consent, as well as fair pay and transparency, arguing that "innovation cannot be used to override artists' rights."

The AI Watchdog database remains publicly accessible, allowing any artist to search their catalog and see what has been used for training purposes.