Spotify has changed its policy on artificial intelligence. The company has stated that these changes are intended to increase transparency surrounding the use of AI in music, combat spam and prevent unauthorised voice cloning. Consequently, Spotify will now begin marking AI-generated music.
According to a press release published by the streaming service, Spotify will implement the DDEX industry standard for labelling AI content and integrate a filter to detect and block manipulative music distribution tactics. These changes come amid growing criticism of the platform regarding the increase in content generated by AI that masquerades as the work of real people.
Unlike Deezer, which has developed its own system for identifying AI content created by neural networks, Spotify is shifting some of the responsibility to labels and distributors. For example, partners will be required to provide standardised information about the use of AI in track metadata via the DDEX system. This will enable them to specify the details of how artificial intelligence was used to generate vocals and instruments, as well as individual parts or entire compositions.
According to Sam Duboff, Spotify's head of marketing and policy, the use of AI in music falls along a spectrum rather than being clearly divided into “AI content” and “non-AI content”. The new standard will enable more accurate disclosure of information about music.
The company has also stated that cloning other people's voices, creating deepfakes and any other form of voice replication or imitation without the copyright holder's permission is prohibited, and that such content will be removed from the platform. While it is not specified exactly how the presence of fakes will be determined, it appears that this again refers to the DDEX system — rights holders will have to update information about all content provided to the platform.
Although the company is implementing the DDEX standard, it is not yet complete. The streaming service has received confirmation from 15 labels and distributors that they are ready to work with the standard under the new rules.
In addition, Spotify plans to clean up user and artist profiles. This refers to profile mismatches, a common fraudulent practice used to inflate play counts. Users illegally publish songs under the names of other artists using similar names or other methods, and receive payments in their own names.
Spotify management has clarified that it does not oppose the creative and ethical use of artificial intelligence. Charlie Hellmann, the service's vice president and head of music, noted that the goal of the changes being implemented is to protect the platform from abuse, not to restrict artists.
He added that the platform's actions would send a message to the whole industry about the need to label music created using neural networks. This sounds like sarcasm.
Let us know in the comments whether you think Spotify is right to start marking AI-generated music. Do you think AI-generated content poses a threat to many industries, or is it just temporary hype that will die down in a couple of years?
