Spotify Removing More Data, AI Copyrights and Compensation Questions

Here are a few quick stories worth paying attention to in the world of Spotify artist data and AI copyrights and compensation.

Spotify Randomly Decides To Remove Three Years Of Data

Spotify is removing a whole lot of your data from the Spotify for Artists platform. In a blink-and-you'd-miss-it banner at the top of some of the song insights on the platform, Spotify says, "We’re ending support for data from 2018 to 2020. If you’d like to save this information, you can download it by June 30."

It's a curious move, given the company just had its creator-focused presentation outlining all of the new features coming to both Spotify for Artists and the Spotify platform itself. The decision to remove data seems to go along with a downsizing of much of the historic data in the platform since 2015. But we've never really gotten a satisfying explanation as to why Spotify is removing data.

On a short explainer page, Spotify simply justifies the move as helping make it possible for them to bring new analytics features to the Spotify for Artists backend. Spotify doesn't actually explain how removing 3 years worth of everybody's streaming data allows them to bring new features like "custom date ranges on more pages, new engagement stats on your Songs page," and more. Spotify notes that the all-time stream counts for songs will remain.

The big issue is that this is your data and you should be allowed to see it. Sure, you can download all of it in a CSV file for the future — which Spotify encourages you to do — but that's a significant leap from being able to see this historic data in Spotify's dashboard (and one of the reasons we're building RootNote — to make this kind of data visualization a breeze). 

Similar to Spotify mysteriously getting rid of all saves older than 28 days, we're probably not going to get a real explanation for this sudden disappearance of data. But it really underscores how important it is for you to own as much of this information on your own computers as you can.

Grimes Offers 50/50 Split To Anybody Using Her AI Voice, Supreme Court Agrees With 'Human Authorship'

Singer Grimes garnered a fair amount of attention (and praise) when she leaned in to the concept of people creative music with an artificially generated "deep fake" of her voice. Her deal? She won't issue copyright takedowns if the creators agree to split the royalties 50/50 with her. 

In her semi-viral Tweet, Grimes explains the reasoning. For her, it's the same deal she offers any artist she features on. And because she has no label or "legal bindings," as she puts it, there's seemingly not much standing in the way of a pretty clean 50/50 split scenario.

Grimes has also since made a few clarifications around the kinds of songs she's willing to allow out into the world. In a Tweet, she explains that she doesn't want anything crossing lines in terms of violence and sexual content (including no "baby murder" songs, as she says), or any "Nazi anthem" unless it's done in jest "a la The Producers." 

Interestingly enough, the U.S. Supreme Court recently ruled that patent holders must be humans. This came after Stephen Thaler sought patents for devices that his AI platform DABUS created. The Court declined to hear his case after the U.S. Patent and Trademark Office stated that patent law clearly outlines that the creations must come from humans. 

This matters because it goes in line with a similar decision from the U.S. Copyright Office, which declined to give Thaler copyright ownership for AI-generated artwork created by DABUS. The copyright office repeatedly stated that "traditional human authorship" was required for a copyright, and AI-generated creations do not satisfy that requirement. 

It seems likely similar reasoning would apply to AI-generated music containing "deep fake" vocals of other artists — provided no other form of "human authorship" went into the recording.