I’ve been here before. We’ve all been here before.
Technologists create new systems that disrupt existing business models, and by the time the industries react to the impact of new tech, it’s already too late.
Such was the case with the Internet, with mobile, with file sharing and now with AI.
When I started working in tech in the early 90s, few people could foresee how the new interconnectivity wrought by the advent of the internet would literally kill industries, while growing others. Overall, it was a net plus, but physical delivery and dissemination of information and many retail services died (HMV, Tower, Sam Goody and countless others in the music sector alone).
And when Napster launched in 1999, the ossified leadership of the music industry refused to believe that file-sharing networks would destroy their incredibly lucrative business of selling content through the delivery of physical product, until it became obvious that they were being completely obliterated.
The fact is, until the infrastructure of mobile and the internet could support the advent of mass-market music streaming networks – again, brought to you, for better or worse, by the tech industry, the music distribution side of the industry was sliding down a sinkhole of diminishing return, half-hearted attempts of DRM, industry-grown efforts (who can forget the unmitigated disasters that were Pressplay and Sony Connect?).
Today, the disruption of the moment is the advent of AI-based music creation services like Suno and Udio. Within the past 12 months, these services have created technologies that allow anyone to create increasingly high-quality music within seconds – something that threatens to upend the business models relied on by every single person in the music industry. In my mind, it’s probably the biggest disruption of the music industry it’s ever faced.
So what is the music industry doing to meet the challenge of this disruption? As itigation.
The RIAA is suing Suno and Udio on behalf of the labels for “massive copyright infringement” based on their claim that the two companies trained their AI with music from the major labels without permission, which the major labels state is copyright infringement.
And while we all know how well the RIAA’s efforts worked out when they attempted to fight file-sharing by suing individual consumers (not well), my main questions here are: does training AI using existing music fair use (as the services claim) and legal or copyright infringement (as the labels claim) and illegal, and the more important question, will suing Suno and Udio really do anything of substance to neutralize the threat perceived by the labels?
To the first point, copyright infringement or fair use? And the answer is probably both.
Having played extensively with both services, it’s obvious the AI has been trained with major label content – if it wasn’t, its musical output for my requests wouldn’t be so accurate and honestly, so good. AI is only as good as what it is trained on, and based on what I have heard, it’s been trained on well-known music across multiple genres and regions.
Of course, proving the extent to which the AI has been trained with the majors content and to what degree could potentially make the case for both sides of the case.
One side could state that they stripped out the data that comprises the uniqueness of the songs as individual copyrighted works, while the labels can state, as they have, that there are unique chord progressions, rhythms and melodies present in the AI-created song output that mimic their copyrighted works.
Ultimately, I come out on the side of the labels on this one, but the bigger point is so what? The genie is already out of the bottle, regardless of whatever is the outcome of this case.
As always, tech moves a lot faster than the content industry and way faster than government regulation and the courts.
And by the time the court case was filed, the launch and adoption of Suno and Udio was a fait accompli, just like every other advancement in technology the content industries try to wrestle control of after the fact.
And in my opinion, this case, though well intentioned, will solve nothing. At the end of the day, Suno and Udio will likely have to agree to some type of settlement, pay off the labels to some degree and move forward with few repercussions.
The case will not stop AI-based creation of music or the advancement of the technology therein. It just won’t. These cases never do.
Perhaps at the end of the day, if there is a way forward that makes any sense, it will come from an agreement between the tech companies and content owners where the AI-music creation companies pay a percentage of their revenue to the labels, publishers and artists, though it’s going to take a while for that to shake out.
In the meantime, the tech companies are going to move fast and break things and and the content industry will try to litigate clawbacks after the fact, as they always do.
