The AI Music Evolution
A couple of years back, an AI-generated song with vocals reminiscent of Drake surfaced. Infamously dubbed “BBL Drizzy”, it heralded a new epoch in the music industry: one where artistry, likeness, and copyright would serve as the battlegrounds. This development led to a heated feud between the big three labels — Universal Music Group (UMG), Sony Music Entertainment and Warner Records, and AI firms Udio and Suno.
The major labels collectively took legal action against the AI firms on grounds of copyright infringement. Furthermore, they kicked up dust with TikTok over the management of AI-generated content on the platform and devised AI tools to monitor the circulation of their music online.
Interestingly, we are now witnessing a detente between the music industry and AI startups. A largely monetizable path forward has emerged, eerily echoing the existing system beleaguering artists.
News came in recently from Bloomberg that the big three have granted Klay, the music startup that champions ethical AI, exclusive deals. Evidently, Klay is planning a streaming service that facilitates users to remix existing songs in disparate styles, courtesy of a model trained on thousands of licensed songs.
The New Venture: KLAY
Acknowledging their collaboration with Klay, Warner Records announced, “KLAY is not a prompt-based meme generation engine designed to supplant human artists. Rather, it is an entirely new subscription product that will uplift great artists and celebrate their craft. Within KLAY’s system, fans can mold their musical journeys in new ways while ensuring participating artists and songwriters are properly recognized and rewarded.” Words that allayed the fears of many.
Collaborations are emerging between the labels and AI music companies, setting a precedent for severing hostilities. UMG and Udio reached a settlement in October, a gesture that was later echoed by Warner Music. The settlement will foster licensing opportunities for music, and fuel potential revenue streams for artists. However, litigation is still trailing behind Suno.
As per a Financial Times report, the labels are urging for a compensation mechanism that mirrors the dawn of music streaming: micropayments rooted in track plays. Although artists and industry leaders like Taylor Swift have attacked this methodology, arguing that it profits the labels more than the music creators. The specifics of the Klay agreements remain unknown, but with AI-generation, it could become very complicated. Consider, for instance, the remuneration complications stemming from a user-generated remix of a popular song going viral on TikTok with millions of views.
The Murky AI Music Landscape
The dynamics of the AI-generated music ecosystem are somewhat complex. Spotify, for instance, removed 75 million “spammy” tracks in just the past year. One of these tracks was “I Run” by the relatively unknown artist HAVEN, a track which shot to virality courtesy of TikTok. Many misattributed the vocals to the R&B artist Jorja Smith. Ironically, the track garnered a staggering 13 million streams before it was taken down. This has led Spotify to enact a new policy against artist vocal impersonations. Although this policy doesn’t cater to original compositions which sound like established artists, drawing us into conversations surrounding a person’s right of publicity.
The track creators ended up confessing that they had written and produced the song, but had processed the vocals using Suno’s AI tool. The tool uses text prompts to generate songs. Following Spotify’s removal, HAVEN. reuploaded the song, this time with human vocals instead of the AI-rendered Smith soundalike. It’s fascinating that many listeners voiced a preference for the AI version.
Such instances paint a surrealistic picture of the music industry’s future. AI-generated tracks wrongly attributed to human artists without any licensing agreement will continue to flood the internet, with the labels constantly seeking to have them removed. Officially licensed AI tracks courtesy of Klay and the big three labels will coexist alongside these unapproved AI tracks. This whirlwind of uploads, removals, reuploads, and modifications inevitably raises convoluted questions about ownership and remuneration. Through their collaborations with AI music firms, the labels are trying to navigate an increasingly nebulous territory: AI music that mirrors our artists is welcome, so long as we are compensated for it.
