Dec 18, 2023 2 min read

Generative AI being used to create fake pre-release ‘pirate’ tracks, says UMG content protection boss

Universal Music’s VP Of Global Content Protection has made a submission to WIPO on the impact of generative AI on music copyright, noting that - among other things - people involved in pre-release leaks are being scammed by people using AI to generate vocal clones

Generative AI being used to create fake pre-release ‘pirate’ tracks, says UMG content protection boss

Generative AI is not only powering a new kind of music piracy, it's also being used to scam people involved in music piracy. Or so says Universal Music's VP Of Global Content Protection, Graeme Grant, in a new submission to the World Intellectual Property Organisation.

"Increasingly fraudsters are using AI to claim they have pre-release tracks which they then make available for sale", Grant notes in his submission. Basically, scammers are using AI tools to create musical clips with cloned vocals that imitate popular artists. They then join online groups where people source and share pre-release tracks and claim that these clips are the real deal, illicitly sourced from artists or labels through hacking or phishing.

"Some people apparently fall for this scam", Grant goes on, "paying thousands of dollars to get their hands on fresh tracks, not knowing that they’re actually faked. Believing these tracks to be authentic, users often engage in ‘group buys’, pooling their resources to meet the fraudster’s inflated asking price, which can range from $5000 to $30,000".

Given that the record companies have been trying for years to crack down on groups that source and share pre-release tracks online, stopping this particular scam presumably isn't a top priority for the music industry. But it's an interesting outcome of the recent developments in music-making generative AI.

What is more of a priority for Universal Music is stopping tracks with AI-generated cloned vocals from being distributed to the streaming services.

Once on those platforms, such tracks can grab a portion of the royalty pool if they gain any traction, either because fans mistakenly believe it's a new track from the cloned artist, or maybe just because they are curious to hear what the AI version of that artist sounds like.

Labels might try to get those tracks removed on copyright grounds - because the AI tool was clearly trained with unlicensed music, or sometimes because the finished track outright samples an existing recording or interpolates an existing song without permission. Other times alternative legal arguments are presented to justify getting a track taken down.

Explaining how Universal sought to get the digital platforms to remove Ghostwriter's fake Drake track - perhaps the highest profile vocal clone release of the year - Grant writes: "The original track contained a sample from a UMG-controlled track called 'No Complaints' by Metro Boomin, [so] was removed on the basis of copyright infringement”.

“A new version ... was then uploaded to [streaming services] with the Metro Boomin sample removed”, he goes on, “which was reported on the basis of trademark and name, image and likeness violations".

Grant has made his submission to WIPO ahead of a meeting next month of the organisation's Advisory Committee On Enforcement which will consider AI.

Elsewhere the Universal exec also outlines the music industry’s biggest beef with generative AI, which is that AI models are being trained with existing music without licence, infringing the major's copyrights.

Existing copyright law should be able to deal with these infringements, he then argues, providing data mining exceptions to copyright aren't introduced or interpreted too widely so that AI companies can legally make use of existing works without permission.

However, he adds, given that labels are sometimes relying on publicity or personality rights to get AI-generated tracks taken down, "additional protection of personal rights (ie voice and likeness) may be necessary".

Great! You’ve successfully signed up.
Welcome back! You've successfully signed in.
You've successfully subscribed to CMU.
Your link has expired.
Success! Check your email for magic link to sign-in.
Success! Your billing info has been updated.
Your billing was not updated.
Privacy Policy