Apple Music is adding AI transparency tags. Independent artists have homework.

Apple Music has confirmed that AI transparency tags are coming, with the responsibility for supplying accurate tags falling on labels and distributors (RouteNote).

If you are signed, that sentence describes someone else’s job. If you are independent, “labels and distributors” means the distributor you upload through — and in practice, the metadata you hand them.

Why this is different from the other announcements

Most AI-labelling news has been about detection: platforms working out, after the fact, what was machine-generated. Deezer has been tagging publicly for the longest and reports AI at roughly 39% of daily uploads, with over 13.4 million tracks detected during 2025 (iMusician).

Apple’s approach inverts it. Rather than inferring, it asks the supply chain to declare. That is a more workable system — inference at that volume is expensive and error-prone — but it moves the burden onto the person uploading.

What that means in practice

Find out what your distributor is doing about it. Most will add an AI disclosure field to their upload flow, if they have not already. Some will apply a default. Knowing which, before your next release, is the difference between declaring and being assigned.

Decide what you are declaring, precisely. The distinction every policy turns on is fully-generated versus assisted, and most real work sits between them. Have an answer you can defend: what was generated, what was written, what was directed, what was rejected.

Do it for the back catalogue too. Tagging a new release is straightforward. Retrofitting an existing catalogue is the awkward part, and it is easier to do deliberately now than in response to a reclassification later.

The thing worth noticing

There is a version of this news that reads as a threat to AI-native artists, and I do not think that is right.

A tag is only damaging if it tells listeners something you were not telling them. For a project that has said 100% AI-generated since the first release, an Apple tag saying the same thing is not an exposure — it is Apple agreeing with the artist bio.

The uncomfortable position is the middle one: artists who used generative tools substantially, described it vaguely or not at all, and will now have a platform describe it for them. That is not a labelling problem. It is a disclosure problem that labelling makes visible.

What I would actually do this week

Open your distributor’s dashboard and look for the AI disclosure field. If it exists, fill it in accurately. If it does not, ask when it will.

That is genuinely the whole task. It takes about ten minutes and it is the difference between being described accurately and being described by a default setting someone else chose.