News · 2026-06-29
Tidal will stop paying royalties on fully AI-made songs
Tidal, the music streaming service, has announced that it will stop paying royalties on songs it identifies as wholly AI-generated. The change takes effect July 15, and it makes Tidal the first major streamer to move past labeling AI music and start actively withholding money from it.
The background: generative AI can now produce a finished-sounding song - melody, instruments, vocals, lyrics - from a text prompt in under a minute, for free or close to it. That has flooded streaming platforms with an enormous volume of machine-made tracks, some uploaded by people gaming the royalty system, since every stream pays a tiny sliver of money. With human-made and AI-made songs sitting side by side in the same catalog and the same payout pool, the question of who deserves the money has gotten sharp.
What Tidal actually did is narrower than 'banning AI music,' and the distinction matters. The company defined 'wholly AI-generated' as a track where every single component was created with generative AI - no human songwriting, no human performance, nothing. Those tracks won't earn royalties. But Tidal is explicitly not removing them from the platform, and it isn't touching music where a human used AI as a tool somewhere in the process. The stated principle is that royalties should 'go to original works directly produced, written, and performed by people,' while still leaving listeners free to play whatever they want. To tell the two apart, Tidal is working with an outside detection partner.
Here's an analogy for how the policy works. Imagine a bookstore that pays authors a small fee every time someone reads one of their books in the store's reading nook. Then a machine starts churning out thousands of auto-generated books overnight, all competing for the same reading-nook traffic and the same fee pool. The store's response isn't to burn the machine books - customers can still flip through them if they like - it's to stop cutting royalty checks for books no human wrote. That's Tidal's move: demonetize, don't delete.
Why it matters: this is a bellwether for how the streaming economy absorbs the generative-audio wave. Spotify, by contrast, has focused on labeling and filtering AI tracks while continuing to monetize them, and has even leaned into AI by letting fans generate covers and remixes. Tidal is staking out the opposite end - that fully synthetic music simply shouldn't draw from the pool meant for human artists. If the detection works at scale, other platforms will face pressure to follow. The reaction on the big tech discussion forum Hacker News was broadly approving, with many users framing it as a sensible dam against a rising tide of low-effort 'slop' that makes genuinely human music harder to find.
The honest caveat is enforcement. Detecting whether a song was 'wholly' AI-made is genuinely hard, and the messy middle - a human songwriter who used an AI tool to generate a backing track, or a producer who cleaned up an AI vocal - is exactly where a blunt detector will make mistakes, potentially penalizing legitimate artists who use AI the way they'd use any other studio tool. There are already signs of the gap: AI 'artists' with millions of streams reportedly remain on the platform without clear AI labels even after the announcement. And as critics note, withholding money for being AI-made isn't really an attribution or copyright principle - it's a quality-and-spam lever wearing the costume of one. Whether that lever is fair, or just expedient, is the debate this kicks off.