August 2, 2024 – In a surprising development, music generation startup Suno has admitted to training its AI model using copyrighted songs, a fact revealed in a court filing on Thursday amidst an ongoing lawsuit filed by the Recording Industry Association of America (RIAA). The lawsuit, which also targets another startup, Udio, was filed on June 24, accusing both companies of using copyrighted music without permission to train their AI models.
Suno’s Admission and Defense
Suno’s recent court admission marks a significant turn in the case. The company acknowledged that it trained its models using recordings, some of which are owned by the plaintiffs, including major music labels. This acknowledgment contradicts earlier suggestions from Suno’s investors that the startup had avoided using copyrighted material without permission.
In the legal filing, Suno stated: “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 Plaintiffs in this case.” This admission came alongside a blog post by Suno’s CEO and co-founder, Mikey Shulman, who defended the company’s practices by likening the training process to a musician learning by listening to music available on the internet.
Shulman argued, “We train our models on medium- and high-quality music we can find on the open internet… Much of the open internet indeed contains copyrighted materials, and some of it is owned by major record labels.” He further compared the process to a young artist inspired by their favorite genre, asserting, “Learning is not infringing. It never has been, and it is not now.”
RIAA’s Response
The RIAA quickly responded to Suno’s filing, criticizing the company’s admission as a forced concession only made under legal pressure. The RIAA strongly rejected Suno’s fair use claim, stating, “It’s a major concession of facts they spent months trying to hide and acknowledged only when forced by a lawsuit. Their industrial-scale infringement does not qualify as ‘fair use’. There’s nothing fair about stealing an artist’s life’s work, extracting its core value, and repackaging it to compete directly with the originals.”
The RIAA emphasized that the core issue was not just about legal semantics but about protecting artists’ livelihoods. “Their vision of the ‘future of music’ is apparently one in which fans will no longer enjoy music by their favorite artists because those artists can no longer earn a living,” the RIAA stated.
The Fair Use Debate
The lawsuit and Suno’s defense bring to the forefront the ongoing debate around fair use in the age of artificial intelligence. Suno’s argument hinges on the fair use doctrine, which allows limited use of copyrighted material without permission under specific conditions, such as for commentary, criticism, or parody. However, the RIAA argues that Suno’s extensive use of copyrighted music for training purposes far exceeds the boundaries of fair use.
The case raises important questions about how AI models are trained and the extent to which companies can use copyrighted content without direct authorization. As AI technology continues to evolve, the legal system grapples with applying traditional copyright laws to new technological contexts.
Implications for the Music Industry
The outcome of this lawsuit could have significant implications for the music industry and AI startups alike. If the court sides with the RIAA, it could set a precedent that limits how AI companies use copyrighted content, potentially requiring them to obtain licenses or develop new methods for training AI without infringing on copyrights.
Conversely, if Suno successfully defends its practices under fair use, it could open the door for broader applications of AI in music generation and other creative fields, albeit with the potential risk of undermining artists’ rights and revenue streams.
Looking Ahead
As the case progresses, it will be closely watched by legal experts, AI developers, and music industry stakeholders. The decision will likely influence how AI companies approach model training and copyright considerations in the future.
The intersection of AI and intellectual property law remains a complex and evolving area, and this lawsuit serves as a critical test of how existing legal frameworks adapt to the challenges posed by new technologies.