Anger at Spotify 'burying' information about AI music
Music creators are set to lose a joint $10 billion (£7.4 billion) worth of income by 2028 because of artificial intelligence
Music streaming platform Spotify must do better to label music that is AI generated instead of 'burying' it away from users, critics told LBC.
Listen to this article
The chair of Parliament’s Culture, Media and Sport committee has warned without greater transparency users could ‘inadvertently support’ music based on stolen work.
Music creators are set to lose a joint $10 billion (£7.4 billion) worth of income by 2028 because of artificial intelligence, according to a report from the International Confederation of Societies of Authors and Composers.
In September, Spotify which is the world’s most popular audio streaming service with 276 million subscribers, updated its AI transparency policy, describing it as 'high impact'.
Read more: Spotify founder Daniel Ek to step aside as chief executive
It will include AI disclosures in song credits, to identify if artificial intelligence has contributed to the vocals, instrumentation or post-production of a track.
But critics say this policy would be ineffective, storing information in hard-to-find places, and as a result very few users would see it.
LBC also spoke to music executives who described it as only a 'small step in the right direction'.
While Spotify does not publish its own data, competitor Deezer said about 900,000 AI songs are uploaded to its own site every month (28% of all uploads).
Chair of Parliament’s Culture, Media and Sport Committee Dame Caroline Dinenage told LBC Spotify has a responsibility to ‘go over and above’ to demonstrate to listeners when they are hearing AI-generated music.
She said: “It’s that fundamental issue of people being paid for what they make and what they produce.
“Nothing about [AI] is original in its purest sense because it's all been scraped from other people's content. You saw that with the Velvet Sundown, the band that gained over 900,000 monthly listeners on Spotify before they finally admitted that their music was AI generated.
The music generator that was used to produce them is being sued by record labels in the US because they've just had this mass copyright infringement.
“It's that fundamental issue of people being paid for what they make and what they produce. Artists are losing revenue, and these are artists that have spent their whole life honing their skills, playing small gigs, playing in their bedrooms, generating their own music, only for it to be nicked by someone else and turned into a nice little cash cow.”
Dr Jo Twist OBE, CEO of UK record labels association the BPI, said: “AI can be an incredible creative tool for musicians when used responsibly and respectfully, but its potential needs to be a realised in a way that benefits human creativity. “Central to this is transparency - both in terms of the music that is being used to train AI models, and the AI-generated music that is increasingly appearing on streaming platforms. It’s vital that streaming platforms make clear labelling a priority - music fans have a right to know where the music they listen to has come from."
Ed Newton-Rex, Founder of Fairly Trained, told LBC: “A huge amount of AI generated music is making its way onto streaming services and diluting royalty payouts for real human musicians.
“[Spotify] say they'll label AI music, but if you read the fine print, that's not what they're doing at all. The first issue is that they're going to rely on the people uploading music, disclosing that it's AI generated. But most people just aren't going to do that.
“They should instead be automatically scanning uploads for AI music and labelling what they find. And then the second issue is that this isn't going to be some label that's easily visible to the user. It's going to be buried in a song's metadata, the niche information that no one ever reads
“As a musician myself, this is really disappointing. It gives the impression that Spotify isn't really interested anymore in being the home of human musicians, which I think is a massive shame.”
Pressure is growing on the UK government to regulate AI content, with ministers expected to bring forward a comprehensive artificial intelligence bill next year.
A spokesperson for Spotify said: “Over the past year, we’ve removed more than 75 million spammy tracks. And during the recent explosion of generative AI tools, we’ve doubled down on our detection systems and policy enforcement.
“We’re taking practical, high-impact steps aimed at making Spotify a more transparent, fair, and trustworthy platform for artists and listeners.
“At the same time, we recognise this is the start of a longer journey as the technology is rapidly evolving. We want to be careful to ensure we’re not penalising the wrong uploaders, so we’ll be rolling the system out conservatively over the coming months and continue to add new signals to the system as new schemes emerge. Developing a standard that works for everyone can take time, but over the coming months, AI disclosures in credits will begin appearing as labels and distributors submit standardised data.
“You can also visit our For The Record page, for more information on this update."