Sony Music Group Hits Pause On AI Training With Content Licensing Push

Artificial intelligence (AI) is constantly evolving, with new applications and capabilities emerging seemingly every day. Machine learning, a core tenet of AI, relies on vast amounts of data to train algorithms. This data can include text, images, and even audio – and that’s where a potential snag has emerged in the relationship between the music industry and AI development.

One of the biggest music groups, Sony Music Group (SMG), recently sent out a wave of cautionary notices. More than 700 companies involved in generative AI and music streaming services received letters from SMG. The letters outlined a new policy regarding the use of Sony Music content in training AI models.

Sony Music Group’s new licensing requirements

The crux of the matter lies in how SMG views the use of its vast library of music, which includes recordings by artists like Beyoncé, Adele, and Harry Styles. The company asserts that its content, encompassing everything from audio recordings to lyrics and album artwork, is intellectual property. As such, SMG is taking steps to ensure control over how this content is used, particularly when it comes to training AI systems.

The letters sent by SMG clearly state that the company prohibits the use of its content in “training, development or commercialization of AI systems” without explicit permission. This extends beyond simply using the music itself. Activities like text and data mining of lyrics, web scraping of artist information, and any other method of extracting or utilizing SMG’s content for AI development are now off-limits without a licensing agreement.

Why the change?

Several factors likely contributed to SMG’s decision to implement stricter licensing requirements. One key concern is the issue of compensation. When AI models are trained on music by artists signed to SMG, the artists themselves are not necessarily compensated for this use of their work. SMG likely views this as a missed opportunity, and by requiring licensing agreements, the company can establish a system where artists are fairly compensated for the role their music plays in AI development.

Another concern is the potential for misuse. AI has the capability to generate new music based on existing works. SMG might be wary of unauthorized AI models creating derivative content that infringes on the copyrights of its artists. By controlling how its content is used in training, SMG can help mitigate this risk.

The move by SMG also highlights a broader debate within the creative industries about the use of copyrighted material in AI development. Writers, filmmakers, and other creators are increasingly raising questions about how their work is being used to train AI models, and whether they are being adequately compensated for this use.

Featured image credit: Possessed Photography/Unsplash