TikTok videos carrying the 'fake Drake' track last week, like the above, nodded to the potential copyright illegality of the recording
MBW Reacts is a series of analytical commentaries from Music Business Worldwide written in response to major recent entertainment events or news stories. Only MBW+ subscribers have unlimited access to these articles.
Yesterday (April 26), the – ahem – ‘artist’ known as ‘ghostwriter’ returned to TikTok with yet another track featuring AI-copied voices of famous people.
Now, this ‘ghostwriter’ isn’t the same ‘ghostwriter’ whose now-infamous Heart On My Sleeve track, featuring a cloned-Drake vocal, caused a global kerfuffle in the music biz last week.
Our more recent ‘ghostwriter’ has a slightly different TikTok handle (ghostwrider777 vs. the original’s ghostwriter977), and refers to his predecessor as “my brother” (see below).
The result of both ‘ghostwriter’ endeavors, though, is much the same: A fake superstar duet – this time of Bad Bunny and Rihanna – created via AI-rip-off vocals, accompanied by a berk bobbing around on camera, face covered by his nan’s net curtains.
The fake Bad Bunny track, Por Qué, is at the time of typing still up on TikTok (where it has over 1.5 million plays), as well as SoundCloud and YouTube. (Interestingly, it has failed to appear on the likes of Spotify and Apple Music.)
As destiny would have it, a few hours after Por Qué arrived on TikTok, Universal Music Group held its quarterly Q1 2023 earnings call – and the major didn’t pull any punches on the subject of its artists being ripped off by generative AI.
Within minutes of his opening remarks to analysts, Sir Lucian Grainge, CEO & Chairman of Universal Music Group, criticized the “content oversupply” that currently sees around 100,000 tracks a day distributed to music streaming services.
“Not many people realize that AI has already been a major contributor to this content oversupply,” said Grainge. “Most of this AI content on DSPs comes from the prior generation of AI, a technology that is not trained on copyrighted IP and that produces very poor quality output with virtually no consumer appeal.
“the recent explosive development in generative AI will, if left unchecked, both increase the flood of unwanted content hosted on platforms and create rights issues with respect to existing copyright law in the US and other countries.”
Sir Lucian Grainge, Universal Music Group
“However, the recent explosive development in generative AI will, if left unchecked, both increase the flood of unwanted content hosted on platforms and create rights issues with respect to existing copyright law in the US and other countries – as well as laws governing trademark, name and likeness, voice impersonation and the right of publicity.
“Further, we have provisions in our commercial contracts that provide additional protections.”
Added Grainge: “Unlike its predecessors, much of the latest generative AI [i.e. ‘fake Drake’ and ‘fake Bad Bunny’] is trained on copyrighted material, which clearly violates artists’ and labels’ rights and will put platforms completely at odds with the partnerships with us and our artists and the ones that drive success.
“Should platforms traffic in this kind of music, they would face the additional responsibility of addressing a huge volume of infringing AI-generated content.
“Any way you look at it, it’s over-supply. Whether or not AI creates [that over-supply], it’s simply bad, bad for artists, bad for fans, and bad for the platforms themselves.”
Grainge’s assertion of the illegality of AI-created music that has been “trained” using copyrighted recordings is a potentially contentious one.
UMG succeeded in getting ghostwriter’s ‘fake Drake’ track deleted by major DSPs via a copyright complaint. But, as MBW was first to point out, ghostwriter may have made a fateful mistake in copyright terms: Heart On My Sleeve appeared to be an original composition, but it included a sample of producer Metro Boomin’s tag (“If young Metro don’t trust you, I’m gon’ shoot you”).
Had that sample not been included on the ‘fake Drake’ track, proving a clear copyright violation – especially in terms of copyrighted recordings from real Drake being used as “derivative” inspiration for the voice of ‘fake Drake’ – may arguably have proven more difficult, and required novel legal arguments.
As Nick Eziefula, partner at UK legal house Simkins, noted in his MBW Views piece the other day: “An AI-powered approach to music-making could herald the next generation of legal disputes, where familiar bones of contention are fleshed out in new ways.”
Universal, though, is very confident that, when it comes to superstar-mimicking vocals created via AI toolkits, the law is already on its side.
On UMG’s Q1 call, Michael Nash, EVP and Chief Digital Officer at Universal (and someone who has written perceptively about AI in music in the past), was asked specifically about his company’s view of the legal framework surrounding copyright protection for superstar-mimicking AI voice technology.
Nash’s answer to that question (from Morgan Stanley’s Omar Sheikh) was unequivocal: ‘Fake Drake’, and other recordings like it, contravene copyright law.
Said Nash: “I’m glad that you asked [this] question about our legal view of AI, because I do think that there’s been a little bit of confusion… we’re happy to have the opportunity to be very, very clear about our view of the legal landscape.
“Companies have to obtain permission and execute a license to use copyrighted content for AI training or other purposes, and we’re committed to maintaining these legal principles.”
Michael Nash, Universal Music Group
“First of all, in terms of copyright, to reiterate our very clearly articulated position – and echo Lucian’s excellent summary earlier – sophisticated generative AI that’s enabled by large language models, which is trained on our intellectual property, violates copyright law in several ways.
“Companies have to obtain permission and execute a license to use copyrighted content for AI training or other purposes, and we’re committed to maintaining these legal principles.”
Nash argued that, in Universal’s view, copyright laws in multiple international territories are already “fit for purpose” on this issue and don’t require an update; i.e. such laws would deem the ripping off of Drake’s voice by a generative AI tool as cut-and-dry copyright infringement.
“But I think it’s very important,” added Nash, “that governments around the world interpret and enforce the existing laws correctively – correctly and actively.”
Nash went on: “Copyright covers all training of AI and copyrighted music regardless of the technical means employed.
“We have further protections [from ‘fake Drake’ etc. appearing on the likes of Spotify] beyond copyright laws under our commercial DSP agreements.”
Nash confirmed recent reports that Universal recently “put all of our [streaming] partners on notice regarding their responsibility to ensure that no third party has unauthorized access to their services for the purpose of training generative AI”.
MBW is sure this quote was said by Nash in the spirit of partnership with the likes of Spotify, Apple Music and co. But there’s obviously another interpretation, too.
Let’s take it as read that tracks like heart on my sleeve are indeed infringing on copyrighted material, whenever said material is ‘fed in’ to an AI platform.
If so, Nash and Grainge appear to be saying, then if the ‘creator’ of such tracks gained “unauthorized” access to this copyrighted material via a particular DSP, Universal may deem that DSP at least partly legally responsible for the infringement.
“soundalikes which serve to confuse the public as to the source or origin, or which constitute a commercial appropriation of likeness in the form of a distinctive voice, are all clearly illegal.”
Michael Nash, Universal Music Group
“We know how critical it is to vigorously enforce our rights here and the rights of our artists regarding any such unauthorized training, and that’s why we’ve taken some added steps to be very, very clear with our partners on this issue,” added Nash.
Echoing Grainge’s words, Nash said that, beyond copyright law, Universal believes that the “exploitation” of famous artists’ voices via copycat AI tech is potentially an infringement of “a number of protections, including trademark, name and likeness, voice impersonation, right of publicity, and all these are instruments that can protect our work”.
Reiterating Universal’s control of master rights, Nash added: “With respect to voice, I think there’s been some confusion here, and I want to be really, really clear on this point.
“We [Universal] own all sounds captured on a sound recording. That is, in fact, the very nature of sound recording copyright and ownership. And here too, depending on the instance, we may also employ name and likeness, voice impersonation, right of publicity protections as well.
“Specifically, soundalikes which serve to confuse the public as to the source or origin, or which constitute a commercial appropriation of likeness in the form of a distinctive voice, are all clearly illegal.”
Universal didn’t only bash generative AI on its Q1 earnings call, though – there was room for optimism too.
In particular, Sir Lucian Grainge gave an interesting answer to a question asking whether UMG would one day consider licensing some of its sound recordings to a company like OpenAI (creator of ChatGPT), so that creators could use such a platform legally build new music based on “derivative” copyrights. (Think: Millions of kids making their own ‘fake Drake’ rap tunes – but legally – and Drake getting paid for it.)
“We’re open to, in terms of licensing, any business solution,” said Grainge. “Obviously [within that], you have to respect our artists and the integrity of their work.
“My philosophy for this company has always been [that] we should be, and can be, ‘the hostess with the mostest’.
“So yes, we’re open for business with [AI companies] which are legitimate, which are supportive, and [with] which we can create a partnership for growth.”Music Business Worldwide
