news

한어Русский языкEnglishFrançaisIndonesianSanskrit日本語DeutschPortuguêsΕλληνικάespañolItalianoSuomalainenLatina

IT Home reported on August 2 that music AI company Suno published a blog post yesterday (August 1) in response to the lawsuit filed by the three major record companies, admitting that it had used the record content of these three major record companies to train AI.However, the company believes that this is legal under the doctrine of fair use.

The Recording Industry Association of America filed a lawsuit

On June 24, the Recording Industry Association of America (RIAA) sued two music generation startups, Udio and Suno, accusing the two companies of using copyrighted music to train AI models.

The music comes from three record companies: Universal Music Group, Sony Music and Warner Music.

Suno responded

Suno responded that it used tens of millions of records to train its AI models, some of which were from the three record companies mentioned above. However, according to the fair use principle of US copyright law, such activities fall within the scope of fair use, which means that no license is required. It also accused the record companies of abusing their copyright.

IT Home translated part of the report submitted by Suno as follows:

Copyright law states that if a copy of a protected work is used for back-end technology and is not disclosed to the public or used to create a new product service in the service that ultimately does not infringe, it is fair use. The U.S. Congress enacted the first U.S. copyright law in 1791, and in the 233 years since then, no case has reached a contrary conclusion. The final conclusion is that making "intermediate" copies of protected works is allowed, but not actionable.

Suno CEO and co-founder Mikey Shulman went on to say in a blog post published the same day as the legal filing:

We train our models on moderate to high quality music that can be found on the open internet… Much of the content on the open internet does contain copyrighted material, some of which is owned by major record labels. Training an AI model with data from the “open internet” is no different than “kids writing their own rock songs after listening to rock music.” Learning is not copyright infringement. It never was, and it is not.