As Bloomberg broke the news over the weekend that the major labels are entering into licensing negotiations with controversial generative AI platforms Udio and Suno - which respectively pitch themselves as the “AI music generator” and “AI for music creators” - artists and songwriters are worried that they are about to watch history repeat itself.
When Spotify launched, music creators weren’t in the room as labels negotiated the licensing deals that set out how everyone gets paid whenever music is streamed. Fifteen years later, they’re still fighting for fair compensation. Now, as major labels sit down with AI companies to decide the future of music creation, artists and writers fear they are being locked out again.
Sony, Universal and Warner are now in talks with Udio and Suno about licensing deals that would bring to an end landmark litigation that was launched a year ago. It’s thought the music companies are looking for both fees and equity in the two start-ups as part of any deal.
What’s less clear is what the artists, songwriters and performers whose music has been used to train these AI systems will get - and so far, no one is talking to them.
Phil Kear, Assistant General Secretary at the UK's Musicians' Union, has some big questions for the majors if and when they license their catalogues for AI training. “Will the consent of the music creators be sought? What share of the licensing revenue will they receive, if any?"
While Kear welcomes the fact that, “at last”, the AI firms “are coming to the table to negotiate legal use of copyright works for training their models”, he adds that it is “very concerning for music creators if record labels intend to agree licensing deals on the basis of existing historical contracts without any further consultation with artists, producers, musicians and songwriters”.
Noting the parallel with the early streaming deals, he adds, “it’s over fifteen years since the major labels negotiated initial deals with streaming services like Spotify, and music creators are still trying to obtain their fair share. Sadly, this feels like history repeating itself”.
Tom Gray, the artist and activist behind the #BrokenRecord campaign and Chair of the Ivors Academy sees something even more troubling - that the major label deals could let AI companies swallow the work of music creators whole with no permission needed and no control given.
“In these deals, it appears recordings would be licensed to AI companies to ingest music-makers’ work without their permission and without reference to their personality or author rights”, Gray explains. “These licences appear to not offer creators an ‘opt-in’, an ‘opt-out’ or any control - whatsoever - of their work within AI”.
In fact, says Gray, the same major labels who have backed the Make It Fair campaign that calls for AI companies to respect copyright appear ready to authorise the use of music creators’ work themselves - for the right price. “The same companies who have stated they wish to ‘make it fair’ seem instead to be ‘on the make’”, Gray observes, adding “I hope I’m wrong”.
The fight isn’t just about money, it’s about whether artists, songwriters and other music creators - including session musicians and studio producers - will have any say in how AI uses their work.
Both the Musicians’ Union and the Ivors Academy - who are campaigning on AI issues alongside the FAC, MPG and MMF - insist that all music creators must consent before their music is used to train AI models. Without that consent being sought, creators can’t say no, can’t negotiate terms, and can’t protect their artistic identity from being regurgitated by AI.
The licensing talks between the majors and the two AI companies they are suing in the US courts come as the debate over copyright and AI becomes increasingly tense within US political circles.
To date, both Udio and Suno have been among the AI companies to insist that they don’t need to secure licences when making use of existing content to train their generative AI models, because AI training constitutes fair use under US copyright law.
The music industry obviously strongly disagrees with that argument, which is why the majors together sued both of the music AI start-ups for copyright infringement.
In its response to that litigation, Suno stated, “It is fair use under copyright law to make a copy of a protected work as part of a back-end technological process, invisible to the public, in the service of creating an ultimately non-infringing new product”.
The wider copyright industries have been involved in wide-ranging litigation and lobbying in the US to try to demonstrate that fair use does not apply to AI training.
A recent report from the US Copyright Office said that AI training might be fair use in some scenarios, but that in other cases licences would likely need to be secured. That was seen as a win for the copyright industries, though President Donald Trump then sacked the boss of the Copyright Office, and his team is working on its own AI report that might reach a different conclusion.
But the fact Udio and Suno, despite their past bullishness about the fair use defence, are now in talks with the majors suggests some AI companies, and investors in AI companies, are becoming more cautious. Which means that there might be more willingness to sit down with major rightsholders to see if terms can be agreed for how licensing the training of a music-making AI model might work.
Meanwhile, for the majors, they might reckon that now is the time to try to pressure Udio and Suno into a licensing deal in order to settle the litigation, before Trump and his team throw a spanner in the works and side with the tech giants on the big fair use debate.
Of course, these talks may come to nothing and the lawsuits could proceed. But if deals can be done, that might set a precedent that results in licensing negotiations with other AI companies. Which will likely prompt even stronger calls from artists and songwriters that they be consulted before their music is used to train AI models.