Spotify on Thursday announced a series of updates on AI’s policy, designed to better indicate when AI is used to make music, reduce the spam and make it clearer that unauthorized vocal clones are not allowed at its service.
The company says it will adopt an upcoming industrial standard to identify and label AI music in credits, known as DDEXAnd soon a new spam of music filter will be released to catch more bad actors.
According to the DDEX system, labels, distributors and music collaborators submit standard AI notifications to musical credits. This solution offers detailed information on the use of AI-such as whether used for vocals, organs or post-production, for example.
“We know that the use of AI is going to be a spectrum, with artists and producers incorporating AI into different parts of their creative work flow,” said Sam Duboff, the world leader marketing and Spotify’s policy, in a press briefing on Wednesday. “This model of industry will allow more accurate, separate revelations. It will not force it to watch in a false binary, where a song is either categorically ai or not AI,” he noted.
In the same announcement, Spotify has clarified its AI personalization policies, stating that AI, Deepfakes and any other form of voice copies or counterfeiting are not allowed and will be removed from the platform.
While the DDEX standard is developing, Spotify says it is taking commitments from 15 labels and distributors who plan to adopt technology and consider its move as it could signal to others, it is time to adopt technology.
Because AI tools make it easy for anyone to release music, Spotify also has a new design to reduce the possible spam that is resulting. This autumn, the company will develop a new spam music filter that will try to tackle adverse mail regularly, highlight them and then stop recommending these pieces to users.
TechCrunch event
Francisco
|
27-29 October 2025
“We know that AI has made it easier than ever for bad actors to collect uploading content, to create copies, to use SEO tricks to handle search or recommendation systems … We are fighting for these regularly for years,” Duboff said. “But AI accelerates these issues with more complexity and we know that it requires new types of mitigation.”
The company said it would gradually release the filter to make sure it targets the right signals, then adds more signals over time as the market evolves.
In this regard, Spotify will also work with distributors to deal with something called “profile mismatches”, a design where someone uploads fraudulent music to another artist’s profile to flow services. The company said it hopes to prevent more of them before music will ever go alive.
Despite the changes, Spotify executives stressed that they continue to support the use of AI provided that it is used in a non -foolish way. “We are not here to punish artists for the use of AI authentic and responsible. We hope that the use of artists by AI production tools will allow them to be more creative than ever,” Spotify VP and Music World leader Charlie Hellman said. “But we are here to stop the bad actors who play the system and we can only benefit from all this good side of AI if we protect aggressively from the disadvantage,” he said.
Spotify updates follow a rapid increase in music created by AI across the industry. This summer, a The zone created by AI called Velvet Sundown became viral in his service, leading users to complain that the company is not transparent to highlight AI tracks. Meanwhile, the flow of opponent Deezer has recently shared that about 18% of the music uploaded every day at its service-or more than 20,000 tracks-is now fully created. (Deezer tells us that the number has now increased to over 30,000).
Spotify will not share its own measurements on the subject immediately – but Duboff told reporters that “the reality is that all continuous flow services have almost exactly the same list”.
“People tend to deliver music to all services,” he explained, adding that the uploading of tracks does not mean that one is listening or that AI’s music makes money. “We know that the use of AI is no more and more a binary, but a kind of spectrum in the way artists and producers use it.”
