Content ID Claims on AI Music: What Happened and How We Fixed It

At the end of October Content ID claims on AI music rose sharply. Why it happens, why no music generator can switch it off, and the lower-claim-risk music we added after three rounds of tests.

At the end of October more of you wrote to us about YouTube Content ID claims on tracks made in Every Mood Studio. Several videos on our own channel got them too. Here's what happened, why no music generator can switch it off, and what we changed.

A cozy studio desk at night with headphones, a laptop showing an audio waveform and a window with rain

What happened

Claims came in on generated tracks, mostly soft piano and calm ambient, and they came from several music catalogs. A claim isn't a copyright strike: it doesn't touch the channel, but it can send a video's ad revenue to the claimant. We wrote down what we found in the first article: which tracks get claimed and how to fix a claim.

Why it happens, and why we can't switch it off

Content ID compares the sound and melody of every upload with reference recordings that rights holders have registered. It doesn't check who made a track first, and it doesn't check whether you have permission to use it. It only checks whether two recordings sound alike.

Millions of tracks are uploaded every day, and more and more of them are AI music that someone registered as their own reference. Calm piano and ambient use few notes, slow tempos and familiar chords, so two tracks made independently can land close enough for an automatic match. That's true for every music generator and stock library, not only for us. Neither we nor any other studio can decide which reference recordings YouTube compares against.

What we tested

What we can choose is the music model. So we tested several other music models in three rounds:

Blind listening. The same track descriptions went to every model, across popular niches: sleep piano, ambient, jazz, lo-fi, spa, acoustic guitar, harp, cello, cinematic and Mediterranean. Excerpts were shuffled and rated without knowing which model made them.
Private uploads to YouTube. Test videos with every model's tracks went up as private drafts, so YouTube's copyright check ran on them.

One model stood out: none of its 24 test tracks got a claim, while tracks from our current music got claims in some niches during the same tests. On sound it's a matter of taste: in our blind listening the current music usually sounded richer, especially on piano, harp and cello, and the new one sometimes won, like on lo-fi.

What changed in the studio

1. You choose the music

On the style step, before the music starts, pick Most beautiful (our richest sound, as before) or Lower claim risk (the new model). You can switch on the Music step too, for new tracks and regenerations. If your channel lives on soft piano and ambient and its earnings matter, try the lower-claim-risk music; if sound comes first, keep the most beautiful one. Listen and decide.

2. A free new version after a claim

If a track gets a claim, open the project, go to the Music step and press the shield button on that track. We make a new version for free with the lower-claim-risk music, once per track, for up to about 1 track in 3 of a project. Then render the video again and, on YouTube, upload the new video or cut the claimed part.

We're doing the same with our own claimed videos.

Lower risk isn't zero risk. Claims are decided by YouTube and the claimant, and no one can promise a video will never get one: not us, not any other music generator. Upload as private or scheduled first and look at the Checks step before a video goes public.

We're sorry if your videos were affected. Write to support@everymoodstudio.com with your project link and we'll help.