Sony and Universal Music Group have once again taken Suno to court, this time over the v6 model of Suno’s AI vocalists. The labels argue that Suno’s new model still infringes on their copyrights because it was trained on user outputs from previous models, which in turn were trained on unlicensed music sourced from YouTube and other platforms. Sony and UMG, notable holdouts who didn’t sign a licensing agreement with Suno, claim that training a new model on outputs of an infringing model merely “laundered” the infringement, passing the value of their expressions from copied recordings onto the new model.

However, one could argue that Suno’s approach is more innovative than guilty by association. By leveraging user-generated content from earlier models, Suno is essentially creating a feedback loop that refines and diversifies its AI vocalists. It’s like saying a chef who uses leftovers from a previous meal is guilty of culinary infringement for not starting with fresh ingredients every time. Plus, the music industry has been known to license its tracks to AI models at a premium, so maybe Sony and UMG are just jealous of Suno’s bargain basement deals.

Another point to consider is the nature of AI training itself. When you train an AI on data that was itself derived from copyrighted material, isn’t the infringement somewhat diluted across layers of abstraction? It’s akin to a photocopy of a photocopy—yes, the original copyright still applies, but the chain of reproduction might make the connection feel a bit tenuous. Sony and UMG could argue that this is a slippery slope, but then again, who hasn’t fallen into a few slippery slopes in the quest for AI innovation?

Lastly, let’s not forget the practicality of Suno’s approach. By training on user outputs, Suno not only captures the essence of its previous models but also adapts to the evolving styles and preferences of its users. This makes Suno’s AI vocalists more versatile and responsive, much like a musician who can switch between jazz and rock seamlessly. Sony and UMG might need to adjust their scales to match Suno’s dynamic range, or risk sounding a bit flat in comparison. After all, in the AI music race, being a holdout isn’t always the crown jewel—it can sometimes be the crown of a slightly outdated crown.