Spotify's AI Song Descriptions: A Consent Problem
Lorde publicly rejected Spotify's AI-generated "About the Song" feature this week after it got her own track wrong. Spotify removed the offending text within a day. The episode points to something bigger: the AI consent battle has quietly shifted from how music is made to how it's described to listeners, and independent artists, with thinner online footprints and no megaphone to force a correction, are the ones most exposed.
The Independent Music Brief | July 20, 2026
On July 16, Lorde publicly called out Spotify's "About the Song" feature after it attached inaccurate context to one of her tracks. "I'm going to go out on a limb and say we don't want this," she wrote on Instagram Stories. "Not only is this inaccurate [...] but reducing a song to an AI-generated meaning right at the source feels like it limits free interpretation. At least make it possible for artists to opt out please." Spotify confirmed the specific text had been removed. Weeks earlier, SZA had made her own objection to AI data-scraping, warning artists: "Do not give away your vibranium! Do not train AI with your genius."
"About the Song" sits apart from the training-data debates of the past year because of where it operates. The AI narrates the meaning of existing music, automatically, at the point of play, on the dominant discovery platform. The consent argument the sector has built around training data now needs to extend to the presentation layer, where platforms are generating content about artists and attaching it to the listening screen by default.
The feature scrapes third-party sources and synthesises them into a summary that appears to listeners as the song's official context. It launched in limited beta in February and is still in beta. Lorde got a correction within a day because she has a global audience and the trade press amplifies her posts. An independent artist who finds an inaccurate description on their single has no equivalent mechanism, no press, no megaphone, no artist-facing correction tool. The only functioning appeals process is fame.
The feature also doesn't distribute its errors evenly. A superstar has years of interviews, press profiles, and annotations for the AI to draw from. An independent artist with a thin online footprint gives the model little accurate material, so it reaches further and misattributes more confidently. The artists most likely to be misrepresented are the ones least able to challenge it.
Lorde's phrase "limits free interpretation" identifies a harm beyond accuracy. A pre-loaded machine summary handed to a listener at the moment of play forecloses interpretation before it can begin. Plenty of independent work earns its audience through ambiguity, music that resists a one-line explanation and means different things to different listeners. A platform that flattens that into a confident AI-authored summary is standardising understanding, and independent artists tend to have more at stake in that gap than most.
The incentives all point toward this feature expanding. It costs Spotify very little to run, it increases time spent in the app, and each artist who encounters a problem confronts it alone. Spotify resolved the Lorde episode by quietly deleting one description, with no policy change attached. The economics and the mechanics don't self-correct. Collective pressure or regulation would shift a default-on, near-free feature like this, and neither kicks in because one superstar got her paragraph removed.
"About the Song" isn't a hostile idea. Context can deepen the connection between an artist and a new listener. But a tool that scrapes unverified sources, synthesises them into confident prose, attaches that prose to an artist's work without consent, and offers no correction path is automated editorialising with the artist's name on it. The value of the feature depends entirely on guardrails the current version hasn't built. The idea could serve discovery. The current implementation works against independent artists. The difference is the design.
The sector can push for a short list of conditions for any AI layer placed on top of artists' releases: a genuine opt-out any artist can use, a correction mechanism that doesn't require going viral, and transparency about the sources being scraped. The sector's leverage is greatest now, while the feature is still in beta and its norms are unset. The task is to convert Lorde's instinctive objection into a written expectation, that anything a platform's AI generates and attaches to a release is subject to the artist's consent, the same principle the sector has spent a year establishing one layer down.
Today's Radar
Apple Music has raised U.S. subscription prices
Individual plans climb from $10.99 to $11.99 and Family from $16.99 to $19.99, roughly an 18% increase on the family tier. Apple attributed the hike to rising licensing costs, following comparable 2026 increases from Spotify, Amazon Music, and YouTube Music. A higher subscription price doesn't automatically mean higher independent income. The total revenue pool grows, but it's shared pro-rata, meaning the overwhelming majority of any increase flows to catalogs with the overwhelming majority of streams. "Rising licensing costs" also points to major-label minimum guarantees that independent labels don't enjoy. The useful thing to watch is whether serial price hikes push subscriber growth into decline in mature markets, and whether any of the increase reaches independent artists in proportion to what listeners are now paying.
VuHaus Group has added KCSN "The SoCal Sound" in Los Angeles
The station's live sessions will now surface on NPR's Live Sessions platform, expanding a human-curated, non-commercial discovery channel in one of the largest U.S. markets. In the same week Spotify's AI began auto-generating song meanings across its catalog, a public-media network expanded a human-curated, non-commercial model in one of the largest U.S. markets. These channels operate outside the engagement-maximising, pay-to-play logic of the major platforms and select music on editorial conviction. As commercial platforms fold more of the listening experience into automated systems, public radio remains a discovery surface where an artist is chosen by a person who decided the music was worth hearing.
Sources
The Hollywood Reporter — Lorde
Digital Music News — Apple Music