Artificial Intelligence
Drake, The Weeknd and Hundreds of Artists Found in Massive AI Music Training Database Leak
According to the report, Drake had nearly 800 entries across the datasets, while The Weeknd appeared in more than 460. Justin Bieber reportedly had 458 entries, with other well-known artists also represented in significant numbers.
A newly released database is raising fresh questions about how artificial intelligence companies collect and use music, after thousands of songs by major artists were reportedly identified in datasets used to train AI music-generation models. Among the artists listed are Drake, The Weeknd, Justin Bieber, Nelly Furtado, Alanis Morissette, and Tate McRae.
The database, published as part of a report by The Atlantic, allows musicians to search whether their work appears in several large music collections that have circulated within the AI development community. The findings have sparked concern among artists who say they never permitted their music to be used in AI training systems.
Major Artists Appear in AI Datasets
According to the report, Drake had nearly 800 entries across the datasets, while The Weeknd appeared in more than 460. Justin Bieber reportedly had 458 entries, with other well-known artists also represented in significant numbers.
The datasets contain millions of songs and have reportedly been downloaded thousands of times by developers and researchers working on AI music-generation technologies.
While the presence of a song in a dataset does not automatically confirm how it was used, many artists argue that their work was included without consent, compensation, or licensing agreements.
Independent Artists Speak Out
Several independent musicians publicly reacted after discovering their music in the database.
Canadian rapper and producer Tre Mission expressed frustration, saying he never approved the use of his music and strongly opposes AI-generated music. Montreal producer Lunice also discovered dozens of entries associated with his catalog, describing the findings as surprising.
Meanwhile, Polaris Prize-winning rapper Backxwash summed up her reaction in a brief social media post: “I don’t like this.”
The revelations have intensified calls for greater transparency around how AI companies gather training data.
Growing Debate Over AI and Copyright
The music industry has increasingly pushed back against AI developers over copyright concerns.
Many artists and rights holders argue that AI systems should not be allowed to learn from copyrighted material without permission. Technology companies, however, often contend that publicly available online content can be used under existing legal frameworks.
The debate has become especially significant as AI-generated songs grow more sophisticated and commercially viable.
AI Music Platforms Under Scrutiny
One company frequently mentioned in discussions about AI-generated music is Suno AI, a platform that allows users to create songs using text prompts.
Court filings previously revealed that Suno trained its models on a vast collection of downloadable music files. Similar concerns have also been raised about earlier projects developed by OpenAI and Google AI, which reportedly used millions of songs to train music-generation systems.
As AI continues to reshape the entertainment landscape, the leaked database may become a focal point in future discussions about artist rights, licensing practices, and the legal boundaries of machine learning.
For many musicians, the issue is no longer theoretical. The ability to search and discover their songs inside AI training collections has transformed a broader industry concern into a personal one.
