In a surprising legal maneuver, music AI startups Udio and Suno have admitted to scraping copyrighted material to train their models. This admission comes in response to a lawsuit filed by major music labels, including Universal Music Group, Warner Music Group, and Sony Music Group.
The labels allege that the startups’ AI models were trained on vast amounts of copyrighted music obtained illegally from the internet. In their court filings, Udio and Suno acknowledged that their models had indeed been trained on copyrighted material. However, they argued that their actions fell under the doctrine of fair use.
According to the startups, the scraping of copyrighted material was necessary for the development of their AI-powered music generation tools. They claimed that the process was a “back-end technological process” that did not involve the creation of direct copies of the original music.
The Recording Industry Association of America (RIAA), which initiated the lawsuit, strongly disputed the startups’ claims. The RIAA argued that the scale of the copyright infringement was far too extensive to be considered fair use. They emphasized that the startups had a legal path to obtain the necessary data through licensing agreements but chose to ignore that option.
As the legal battle unfolds, it remains to be seen whether the startups will be able to successfully defend their actions under the fair use doctrine. The outcome of this case could have significant implications for the future of AI-generated music and the rights of music creators.