AI vs AI: New Labeling Rules

AI labeling rules are becoming part of the independent artist’s release process. Before uploading, artists need to know what a tool generated, where that material will appear, and which disclosure that destination requires.

That sounds straightforward until a project includes human vocals, a generated instrumental layer, automated mastering, and an AI-created video. Those contributions raise different questions. One broad statement about “using AI” may not describe the release accurately.

For September 2026, the practical priority is documentation. Keep enough information to explain how the finished work was made, then apply the rules for the platform receiving it.

1. AI labeling rules distinguish identity from production

Spotify announced that AI Persona badges would begin appearing in mid-September 2026. As of September 7, that remains an announced rollout. The badges concern artist identities that may be AI-generated and do not represent real people. Spotify also describes default restrictions on recommendations for those identities, with exceptions and an appeal process. Source: Spotify’s AI Persona announcement.

An artist identity and a production technique are separate matters. A real musician using an automated production tool should not assume that an identity policy automatically describes their situation.

Review the actual scope of the policy. Consider the artist profile, the audio, and the accompanying visuals individually before deciding what must be declared.

2. Check the rules for your upload route

YouTube’s creator guidance requires disclosure for certain realistic or meaningful AI-generated or altered content, including generated music. It distinguishes that from minor production assistance. YouTube also says disclosure itself does not restrict audience reach or monetization eligibility, although other policies still apply. Source: YouTube creator disclosure guidance.

For music delivered through partners, YouTube provides separate categories: Fully Gen AI, Partly Gen AI, and No Gen AI. Its examples distinguish generated musical contributions from ordinary pitch correction and automated mastering. Source: YouTube music-partner guidance.

If a distributor delivers your music, ask how its form maps to the destination’s requirements. The terminology in your dashboard may differ from the terminology in a platform help article.

3. Keep a production record with the master

G.U.M.M.’s recommendation is a short production record containing four fields:

• The tool used.
• The task it performed.
• The generated material retained in the release.
• The disclosure submitted to each destination.

Collect those details from collaborators before final approval. The producer may understand the audio process while the artist or designer understands the visuals. Nobody should have to guess about another person’s contribution.

For example, “AI used” tells a distributor very little. A note explaining that a generated background texture appears in the video, while the recording uses human performances, gives the team something concrete to assess.

Save the approved record alongside the final audio and artwork.

Disclosure does not settle every rights question

A platform disclosure describes how content was produced or presented. It does not independently establish that every sample, voice, image, or other contribution was authorized.

Keep agreements and permissions with the production record. If a collaborator supplied material under a license, retain the relevant terms and the version that applied when the material was obtained.

Similarly, a platform label should not be treated as a copyright decision. Questions about authorship, ownership, permission, distribution eligibility, and audience disclosure may require different answers.

Make disclosure part of release approval

Before delivery, assign one person to compare the production record with the answers entered in the upload form. Save a copy of the submission and any clarification received from support.

Repeat that review when the music, video, artwork, or artist presentation changes. An answer that accurately described an earlier version may not describe the final release.

AI labeling rules will continue developing. Accurate records give independent artists a practical way to respond: document what happened, follow the destination’s instructions, and keep the evidence behind each answer.

Related reading: DSP Declarations & Delivery Changes · Metadata Is Your New Defense