←── back to feed
/topics/atlantic-music-training-database-investigation

Atlantic music training database investigation

2 items2 sourcesupdated 46d agotrend 0

The Atlantic's investigation uncovered four datasets containing millions of songs used to train AI music models, which reporter Alex Reisner made publicly searchable. Two datasets contain 12 million and 9 million tracks respectively, while two smaller sets hold over 100,000 songs each; Google and Stability AI have confirmed using them in research.

  • Four music datasets identified for AI training, ranging from 100,000+ to 12 million tracks
  • Reporter Alex Reisner created searchable public database of the training music
  • Google and Stability AI confirmed downloading and using the datasets in research papers
  • Datasets downloaded thousands of times; exact usage by other companies unknown
  • Free Music Archive among sources included in the training data collections