The Atlantic has launched a publicly accessible, searchable database cataloging music datasets leveraged to train generative artificial intelligence models. Compiled by reporter Alex Reisner, the tool consolidates four distinct music repositories, revealing the sheer scale of copyrighted audio ingested by the industry. Two of the datasets alone contain twelve million and nine million tracks, respectively, while the remaining two each exceed one hundred thousand songs. Investigation reveals that these repositories are primarily distributed as automated link lists targeting YouTube and Spotify. AI development teams employ scraping utilities to harvest audio files, systematically bypassing login gates, advertisements, and creator monetization mechanisms. This practice directly contravenes platform terms of service and raises significant licensing concerns, particularly since sources such as the Free Music Archive restrict commercial exploitation. Despite these restrictions, the datasets have been downloaded thousands of times, with Google and Stability AI explicitly acknowledging their utilization in published research. The catalog encompasses an extensive range of artists, spanning pop icons like Lady Gaga and Fred Again, rock legends including Radiohead and Bruce Springsteen, hip-hop pioneers the Wu-Tang Clan, and experimental composers such as Aphex Twin and Hainbach. By centralizing this information, the database transforms opaque training practices into verifiable public records. It provides journalists, policymakers, and industry stakeholders with concrete data to assess copyright compliance and developer accountability. The initiative underscores the mounting pressure on artificial intelligence firms to address transparency and equitable compensation as model capabilities continue to expand.