Apple Music told record labels this week that visible "Made With AI" tags are coming later this year, forcing the industry to publicly disclose which songs are AI made just as fraud-driven AI streams threaten to bury real artists.
On August 20, Apple sent a message to music industry partners confirming that its AI Transparency Tags, which labels and distributors have been able to attach to uploads since March, will soon show up directly to listeners as "Made With AI" labels. Until now those tags lived only inside Apple's back-end content delivery system, invisible to anyone streaming the song. Not any more. Soon they'll sit right on the track page, according to reporting from Billboard and AppleInsider.
The rule applies whenever AI generated a material portion of a track: the audio, the artwork, the lyrics, or the music video. Apple's own language is broad. Anything "primarily derived from a generative AI service" counts as AI platform generated and has to be tagged. Apple hasn't set an exact launch date. Only that it's coming before the end of 2026.
Nearly Half the Uploads, Almost None of the Streams #
The timing isn't an accident. According to a study by SubmitHub, which built an AI detection tool called SH Labs after noticing what it called a growing disclosure problem, nearly 40% of music released in July 2026 showed some trace of AI involvement. Of that, 23.2% was fully AI generated and another 15.3% was AI generated then reworked by a human. Worse, SubmitHub found that 31% of the songs its tool flagged as AI had been submitted by their own ups as not AI at all.
Deezer's numbers tell a similar story, and they're moving fast. The French streaming service said in April that AI tracks made up 44% of daily uploads, or roughly 75,000 tracks a day. By July that figure had crossed 50%, with close to 90,000 AI generated tracks landing on the platform daily, Deezer reported.
Deezer launched a free AI music detection tool on June 11, 2026 that scans playlists across 20 streaming platforms, including Spotify and Apple Music. With 44% of its daily uploads now AI-generated and 85% of those streams flagged as fraudulent, the company is betting its detection infrastructure is valuable enough to license to rivals and rights... - AI generated music detector for streaming services - how to find AI music in your playlists
Yet almost nobody is actually listening to any of it.
Deezer says AI generated tracks still account for only 1% to 3% of total streams on its platform. The company estimates 85% of those plays are fraudulent: bots streaming AI tracks to farm royalties, not real listeners tuning in. That's the real story hiding under the upload numbers. The flood isn't artists finding an audience with AI tools. It's a scam running on autopilot, and disclosure labels do nothing to stop the bots from playing songs that are already tagged honestly.
Tags Won't Catch a Bot #
Here's the catch. Apple doesn't independently verify any of this. The whole system runs on labels and distributors self-reporting whether a release used AI, the same honor system that let 31% of SubmitHub's flagged tracks go mislabeled in the first place. A tag only works if the people filling it out have a reason to tell the truth, and right now the only reason is Apple asking nicely.
Apple isn't moving alone, either. Spotify rolled out its own AI song credits in beta on April 16, 2026, starting with the distributor DistroKid. It uses DDEX metadata to show a small AI tag inside a track's credits when an artist voluntarily declares that AI touched the vocals, instruments, or production. Voluntary is the key word there. It's still not visible on desktop, and Spotify has said outright that disclosing AI use won't hurt an artist's ranking. Spotify also announced a spam filter last September that has pulled more than 75 million tracks in the past year, aimed less at AI itself and more at the mass-upload behavior that AI makes cheap.
Same instinct, different tool. Apple carries enough weight in the industry that making its tags mandatory and visible puts real pressure on the rest of the market to catch up, the same way iTunes once forced labels to standardize pricing on downloads. YouTube Music has no equivalent visible label yet. If Apple's tags actually stick and listeners start noticing them, that gap becomes hard to defend.
None of this touches the fraud problem directly. A track can be honestly tagged "Made With AI" and still get streamed 90,000 times by bots overnight, and a track that's dishonestly left untagged slides through exactly as it does today. Labeling tells you what a song is made of. It doesn't tell you who, or what, is actually listening to it. That gap is where the real fight over AI music is headed next.
Also read: Lutnick Asks Judge to Toss Lawsuit Over the Government's Intel Stake • LinkedIn Users Click Its AI Slop Reporting Button a Million Times • Amazon Quietly Raised Prices on Echo, Kindle, Fire TV and Eero Devices
LinkedIn Users Click Its AI Slop Reporting Button a Million Times LinkedIn's new "seems like AI slop" button has already been clicked more than a million times in its first two weeks, and flagged posts are now getting 40% fewer views. The rollout follows research from Pangram Labs showing more than 40% of LinkedIn's long-form posts are fully AI-generated, the highest rate among major platforms studied. - how to report AI generated content on LinkedIn - why LinkedIn users hate AI slop posts