Previously, we discussed how streaming services like Spotify and YouTube can overrun with AI Music. Some estimates predict an astounding $4 billion annual annual revenue for “AI artists” by 2028, or up to a quarter of the music streaming revenue. If that's the problem then Spotify thinks so too. A kind of thing.
This month, Spotify revealed that it had removed 75 million trucks from its platform over the past year. The reason was a flood of AI-generated spam and deceptive uploads.
“Spam tactics… AI tools make it easier for everyone to generate a lot of music, making it easier to exploit,” Spotify said in a statement.
AI double-edged sword
Musicians have always embraced technology, from electric guitars to auto tuning. However, the generation AI is different. With a few clicks, anyone can solve endless songs. Some are creative experiments. Mostly it's just noise. Worse, scammers use AI to impersonate real artists, flood algorithms and siphon royalties.
“In the worst case, AI can confuse or deceive listeners by bad actors and content farms,” Spotify said in a policy update.
However, Spotify is not trying to ban AI entirely. On the contrary, executives say they believe that AI can become a powerful creative tool if used responsibly. The real concern is when AI is deployed at scale to exploit systems.
“We're not here to punish artists for authentic and responsible use of AI,” said Spotify's VP and Charlie Helman of Global Head of Music. TechCrunch. “But we're here to stop the bad actors who are betting on the system.”
So, what does Spotify actually do?
High-tech companies generally take very well Wraith Fair Attitude towards AI. As long as the content flows and there's a view or listening flow, everything's going well. Of course, everything doesn't go well. AI content often involves impersonating real people or bands, or creating fictional bands to trick listeners. Spotify's new Crockdown is designed to restore confidence in streaming music.
First, I'm running the music spam filter. The system identifies patterns of abuse (a massive uploads, replica tracks, SEO hacks, or ultra-short audio designed to game games) and tag recommended algorithms before they can be contaminated. Essentially, when you try to game the system, you'll probably get flagged. If flagged, Spotify will stop promoting songs and often remove them.
Secondly, Spotify has strengthened its rules regarding vocal deepfakes. From now on, the voice generated by AI imitating a real artist will only be permitted if the original artist gives explicit permission. This brings us much closer to what music really is.
“Use the misuse of AI to help artists use their identity and undermine their artistry and threaten the fundamental integrity of their work,” the company says.
Spotify is testing a new tool to stop “profile mismatches.” This is the practice of someone uploading fake or other songs in the name of a famous artist to get into fame. This kind of spoofing will be easier with AI, and Spotify says artists will be able to report these issues even before the song is released.
How much is the problem?
In 2023, Spotify quietly changed its royalty rules. Now, before you make money, you need to stream songs at least 1,000 times. This update was partially designed to combat scammers who uploaded thousands of short AI tracks set at the 30-second play threshold of loyalty payouts.
Despite these efforts, the scale of the problem is difficult. According to Spotify, the 75 million spam tracks have never been brought online through filters or have never been removed after the fact. That number is almost three-quarters of the size of an active catalog of 100 million songs. But Spotify claims that this flood of junk hasn't changed the way people actually listen to music.
“The engagement with AI-generated music on our platform is minimal and does not affect human artist streams or revenue distribution in meaningful ways,” the company says. Guardian.
To address the growing confusion about the real and the things that aren't, Spotify is adopting a new credit system called DDEX. This is a high-tech industry standard that allows artists, labels and distributors to clearly state whether and how AI was used to create songs, whether vocals, instruments, or post-production.
“This industry standard allows for more accurate and subtle disclosure,” says Sam Duboff, head of global marketing and policy at Spotify. TechCrunch. “It doesn't track down any false binary that has to be either resolutely AI or at all AI.”
Disclosure is optional, at least for now. But Spotify says they are the first step into a more transparent streaming landscape.
“The change is to strengthen trust across the platform,” Spotify said. “It's not about punishing artists who use AI responsibly.”
The next act of streaming
AI is threatening to take over the music industry completely (among many others).
Spotify's recent moves have conveyed much greater disruption in the music industry and reflected wider tensions. Streaming rival Deather recently revealed that nearly 30,000 trucks (one fifth) uploaded daily are fully generated. The numbers are growing rapidly.
Meanwhile, experimental projects like Velvet Sundown, a self-proclaimed “synthetic music project,” continue to find audiences. Spotify doesn't beat the band because it doesn't break the rules. However, this case promotes public pressure on essential AI labels. On the other hand, fake lofi or instrumental playlists are already common on YouTube.
Spotify executives say they don't create or own music on the platform. But as gatekeepers to hundreds of millions of listeners, their policies shape the industry in a strong way. The current challenge is to balance. Protecting artists, providing information to viewers and preserving room for innovation is without taking over the slop.
