The music streaming service Spotify has unveiled several measures to encourage artists and publishers to be more transparent about their use of artificial intelligence, as well as to limit certain abuses.
The Swedish platform is recommending that musicians and producers comply with a new standard developed by the Digital Data Exchange (DDEX), a consortium of leading media companies, music licensing organisations, digital service providers, and technology firms that develops standards for the creative industries.
Since the beginning of the year, DDEX has allowed tracks to be labelled as entirely, partially, or not at all created with AI in their descriptions.
Once these metadata are integrated, they will be available "across Spotify", promised Sam Duboff, Head of Music Marketing at the streaming platform.
The issue gained prominence in June when an AI group called The Velvet Sundown suddenly went viral, with their most popular song surpassing three million streams on Spotify.
The new labelling system operates on a voluntary basis, and Spotify does not require content uploaders to disclose AI's role in their production.
"Initially, I think, people's mindset was very much binary," explained Charlie Hellman, Head of Music at Spotify, during a presentation.
"There's either AI music or there's not. But the reality is that we're now seeing this proliferation of so many different ways that AI is incorporated into all different steps of the tool chain."
Hellman said Spotify does not want to "punish artists for using AI authentically and responsibly".
According to the company, more than 15 labels and distributors have already committed to comply with the DDEX nomenclature.
Deezer is currently the only major audio platform to systematically flag tracks entirely generated by artificial intelligence.
Regarding such tracks identified by Spotify as entirely created through generative AI, Duboff said "their audience is minimal".
"It's really a small percentage of streams. In general, when the music doesn't take much effort to create, it tends to be low quality and doesn't find an audience."
The platform has also announced that it has updated its rules to make clear that unauthorised AI use, including the creation of deepfakes or imitations without consent, is not permitted and such content will be removed.
Source: AFP
