An investigation by The Atlantic has revealed that millions of songs, including tracks from prominent artists like Taylor Swift and Bad Bunny, have been used to train artificial intelligence models. The publication released four searchable databases containing a total of 12 million songs, 9 million songs, and approximately 100,000 songs in each of the two smaller databases.
According to staff writer Alex Reisner, these databases provide a comprehensive overview of the extent of copyrighted music utilized for AI training. Legal challenges are currently in progress against generative AI music platforms such as Suno and Udio, which are invoking fair use as a defense for using copyrighted material. Reisner noted that a previous legal case in book publishing did not successfully argue copyright infringement, but piracy claims resonated more strongly with the courts. The initial settlement for that case was $1.5 billion, and final results are still pending.
The databases may serve as valuable resources for the music industry as they navigate potential lawsuits related to AI music training. Music streaming services have implemented measures to prevent, identify, or label generative AI creations, although these efforts have seen mixed results. Issues with scammers creating imitation bands through AI also persist, as these entities attempt to profit from unauthorized reproductions of existing artists’ work.
