The Recording Industry Association of America is suing Suno AI and the developer of Udio

The Recording Industry Association of America (RIAA) recently announced it filed a lawsuit in Boston against Suno AI and one in New York against Uncharted Labs, the developer of Udio AI. The plaintiffs include Sony Music Entertainment, UMG Recordings, Inc., and Warner Records, Inc.; practically every company holding the copyright to sound recordings scraped by either of the services, and the lawsuits seek recognition that both services infringed copyrighted sound recordings, injunctions to prevent Suno and Udio to continue infringing copyrighted recordings belonging to the plaintiffs, and damages calculated at up to $150,000 per infringed work.

The main claim supporting both lawsuits is that Suno and Udio have exploited unlicensed copyrighted content for profit; an action that cannot be justified by appealing to fair use doctrines. Moreover, the RIAA has an excellent point that could prove useful to convince judges that there is harm being done without appealing to the "regurgitation of pre-existing content”: now that it is in everyone's reach, synthetic music generation will cheapen and flood the market with quantity, regardless of the quality, which will very likely drown out the works that made the technology possible in the first place.

However, there's proof that contrary to what Suno and Udio would want the general public to believe, their services do produce near-identical replicas of copyrighted works, ranging from "Great Balls of Fire" to Mariah Carey's "All I Want for Christmas Is You". At best, these generations would count as covers of said works, albeit still unlicensed. Moreover, it is impossible that the service would have arrived at those song fragments if the model had not been trained on those very works, to begin with. There is also the fact that at least one person closely related to Suno has not been very discreet about the fact that copyright-infringing scraping secured an investment in the company in the first place.

Long considered the enemy, it seems that the RIAA has a pretty strong case and in an unlikely turn of events, the support of many of us. Both companies have already issued their public and impassioned defenses of their technologies, emphasizing how their companies value originality and their platforms are not intended to replicate copyrighted works (likely an attempt to offload responsibility to the users). Moreover, it is also likely that both companies will attempt to appeal to fair use doctrines to justify their content-scraping endeavors, much like Meta recently claimed that it was collecting personal data under a "legitimate interest" to provide possible service to its European customers.

It is impossible to predict how this saga will end. By now, using vague legal-sounding terms like "fair use" and "legitimate interest" has become commonplace for AI firms looking to evade responsibility and claim that whatever they were caught doing is not as illegal as it seems. However, unlike previous lawsuits regarding scraped content, there is no arguing that for instance, the model heard a 15-second sample of a song and managed to fill out the rest entirely on its own.

More importantly, the RIAA seems poised to demonstrate that copyright material has been misused and that the misuse of copyrighted material is damaging to the original owners' and creators' business, automatically disqualifying the defendants' appeal to fair use. If successful, the RIAA could pull out both services off the market. While Suno or Udio could attempt to continue by training models with licensed and publicly available work, by their own (and their investors') admission, costs would rise, quality would dramatically diminish, and the appeal of the service would vanish.

Thinking optimistically, there could be a lesson on the hubris of generative AI companies regardless of the outcome. However, there is a more realistic sense in which we all know that the RIAA needs to win to drive the point home.