Music AI company Suno has formally responded to the lawsuit filed against it by the three major record companies, admitting that it trained its generative AI model with recordings owned by the majors, before presenting a bombastic fair use defence. It then accuses the record companies of anticompetitive behaviour which, it says, constitutes misuse of copyright.
An accompanying blog post reveals that, before the majors filed their lawsuit last month, Suno was in talks with the record companies about how they could collaborate to “expand the pie for music”. As a result, Suno isn’t sure if the new lawsuit is “the result of over-eager lawyers throwing their weight around” or “a conscious strategy to gain leverage in our commercial discussions”. Either way, “we believe that this lawsuit is an unnecessary impediment to a larger and more valuable future for music”.
In their lawsuit, the majors said that Suno had been vague about what music it had used to train its AI, but that they were certain the training dataset included recordings owned by Universal Music, Sony Music and Warner Music. In its response, filed with the court earlier today, Suno says, “It is no secret that the tens of millions of recordings that Suno’s model was trained on presumably included recordings whose rights are owned” by the major record companies.
The music industry is adamant that if an AI company copies recordings onto a server as part of its training processes, that copying must be licensed. But many AI companies reckon such activity is covered by the fair use principle under American copyright law, meaning no licence is required. And Suno is very certain that fair use applies.
“It is fair use under copyright law to make a copy of a protected work as part of a back-end technological process, invisible to the public, in the service of creating an ultimately non-infringing new product”, its legal filing states.
“Congress enacted the first copyright law in this country in 1791”, it continues. “In the 233 years since, no case has ever - not one single time - reached a contrary conclusion. Every single time the question has been presented - and it has been presented over and over and over again - the ultimate conclusion has been that making an ‘intermediate’ copy of a protected work … is permissible, not actionable”.
And while copying at the input stage of Suno’s AI is fair use - as far as Suno is concerned - there is no copyright infringement on the output, ie the music Suno generates, because the model learns from the recordings in its training dataset, it never samples or adapts them. “Let there be no doubt: the outputs here are, as a rule, non-infringing”, the lawsuit adds.
Convinced that its hands are clean, Suno then proceeds to accuse the major labels of wrong-doing. The major labels “frame their concern as one about ‘copies’ of their recordings made in the process of developing the technology”, it says, “but what the major record labels really don’t want is competition. Where Suno sees musicians, teachers and everyday people using a new tool to create original music, the labels see a threat to their market share”.
Under US law, another defence in copyright infringement cases is misuse, which applies where the copyright owner can be shown to have behaved in an anticompetitive way. And that too, Suno reckons, is grounds for defeating the major record companies’ lawsuit.
“Suno will ultimately prevail in this litigation”, the lawsuit declares, “because the values that copyright law embodies and protects are Suno’s values: incentivising innovation in the service of enabling more expression of more ideas by more people” The majors’ “lawsuit reflects the opposite values: an attempt to misuse intellectual propety rights to shield incumbents from competition and reduce the universe of people who are equipped to create new expression”.
Although Suno’s defence is particularly bombastic, it is pretty much in line with the responses made by other AI businesses that have been sued by companies that own the copyrights in images, newspapers, books and other content. The music industry - and the wider copyright industries - remains certain that AI training is not fair use. Whichever AI case gets to trial first will likely set the precedent regarding the fair use defence.
