Scraping data off the web is massively denounced by companies, especially music labels, but Suno is defending its actions as it is currently facing a lawsuit from three major global labels. The latest lawsuit is against Suno and another music generation platform, Udio, who both allegedly scraped data from songs found on the web for their AI training, to which Suno admitted to.
It was revealed by the music generation platform that songs found on the internet are "fair use," and this is against labels building legal training by obtaining licenses.
Suno Admits to Data Scraping for its AI Training
Suno
According to the latest court filing earlier today, Suno has admitted to resorting to data scraping as part of its technological development, particularly in training its AI. This claim from the company is part of the ongoing lawsuit that it is facing from three major music labels including the Universal Music Group, Warner Music Group, and Sony Music Group.
These three music labels fought against Suno and another AI platform, Udio, regarding their illegal access and use of copyrighted content to train their models.
It is also known that the Recording Industry Association of America pushed for this lawsuit against the two AI companies due to their notorious access to copyrighted data.
In the filing, Suno claimed that these copyrighted songs found on the web are "fair use," especially if the company hides their training and access from the public. Suno said that this clause could be found in the copyright law, citing it in their defense, with making copies as part of a "back-end technological process" as it leads to creating a "non-infringeable" new product.
Data Scraping Issues for AI Training
Unlicensed and unauthorized access to files for AI training has been a massive issue since generative AI was introducedlast year, and there have been significant laws and company copyrights that have targeted perpetrators who are data scraping. Multiple lawsuits have been filed against OpenAI and Microsoft, with one of the stellar cases centering on the New York Times.
On the other hand, some companies traverses thin ice regarding data scraping, with Meta notorious for obtaining user data from its social media platforms for its AI training. However, there is a way to avoid having one's data used for Meta's large language model development, and it is through not using or posting on the likes of Instagram.
Furthermore, there are companies like Reddit that allow the use of its platform's data for AI training, but it is under licensing and its partner would have to pay a substantial amount like Google's $60 million per year. Despite thisgrowing legitimacy of AI training, some companies still resort to scraping, with Suno admitting to it amidst the ongoing lawsuit.
