YouTube Music Just Labeled Primarily-AI Songs. Your Release Stack Needs a Disclosure Plan.

AI disclosure used to be a values debate. As of this week it is a metadata field, and the field ships with every release.

On October 14, 2026, YouTube Music started labeling songs "created primarily using generative AI." The same week, Deezer said it is building a toggle that lets listeners filter AI-generated music out of search. Apple Music, Spotify and TIDAL each have their own version. The question is no longer whether platforms will ask how a record was made. It is whether your answer is accurate, consistent everywhere and ready before a listener, a journalist or a creator asks.

Treat disclosure like artwork and credits: a release asset with an owner.

What YouTube actually shipped

The official announcement is short, and the details matter:

• What gets a label. Songs "created primarily using generative AI." YouTube's example is "using a generative model to create the composition and vocals."
• Where it shows. "When you listen to a track on YouTube Music where AI played a significant role, you'll see a label directly on the player page." Tapping the label opens a menu with more information.
• How YouTube decides. It is "teaming up with music partners and using our own detection tools," applying "the same labeling principles already used across YouTube."
• Where it goes next. The labels "will also expand to the main YouTube app," which The Verge reports will follow later.

Notice what is missing. As Music Business Worldwide points out, neither the blog nor the Help Center page says whether labeled tracks are treated differently in recommendations or royalty payouts. TechCrunch frames YouTube's approach as disclosure rather than blanket restriction: AI involvement alone does not make a track ineligible for monetization. Don't plan around a penalty YouTube hasn't announced. Plan around a label your audience can now see.

Fully, Partly, No: the disclosure taxonomy

The operational detail lives in YouTube's Help Center page for music partners. Partners declare one of three values when they deliver content:

• Fully Gen AI. All content created with generative AI. Examples: a text prompt into an AI music generator that downloads a finished track, or a model that generates the composition, instrumental performance and vocals across multiple prompts.
• Partly Gen AI. Part of the content created with generative AI. Examples: an AI bassline or string section under live vocals and instruments; AI used to brainstorm lyrics or iterate melodies before a studio recording; and a song with 100% human-performed audio whose music video came from a text-to-video model.
• No Gen AI. Examples: standard pitch correction such as Auto-Tune on a live vocal, or automated AI mastering for EQ, compression and balance.

Declarations go through DDEX ERN 4.3.2 or later, CSV delivery templates (a contains_ai value of all, partly, none, or unknown to reset a wrong entry), or the attribute section of YouTube Studio for direct uploads.

Three lines on that page change release ops:

1. Silence is not neutral. YouTube "may use additional signals," and if partners include no Gen AI information, it "will rely solely on those signals and may designate content as fully or partly Gen AI." Skip the field and a detector decides.
2. Interpretation is on you. "It is up to partners to interpret the Gen AI definitions provided by DDEX in relation to their content." Your distributor or label needs your answer, in writing, per track.
3. The visuals count. An AI-generated music video can push an otherwise human recording into "Partly." Your video team and your metadata team now need to talk.

One open question: the blog does not say whether "Partly Gen AI" tracks carry the player label. Plan as if accurate metadata will travel further than today's interface.

The DSP transparency map: same question, different consequences

Every major platform asks roughly the same question. They do very different things with the answer.

• YouTube Music. Labels for primarily-AI songs on the player page, from partner declarations plus YouTube's own detection, expanding to the main app. No announced recommendation or royalty change.
• Apple Music. Per MBW, Apple told partners in August that its AI Transparency Tags would become visible as "Made With AI" labels later this year, dependent on labels and distributors declaring AI-generated content.
• Spotify. Testing AI disclosures in song credits since April via label or distributor. In August it said it would add an "AI Persona" badge to profiles with AI-generated identities and keep that music out of editorial and algorithmic recommendations by default.
• Deezer. Detects and tags AI music with its own technology. This week it said a toggle is coming before year-end that hides AI-generated tracks, albums and related artists from search, citing a survey in which around 45% of its users wanted a way to remove AI music.
• TIDAL. Said in June it would tag fully AI-generated music and stop paying royalties on it.
• The industry frame. In July, the RIAA and IFPI proposed two voluntary labels: "AI-generated" for tracks built entirely by AI or with an AI-produced lead vocal or main instrumental, and "AI-assisted" for mostly human-made tracks with AI used in places.

One record can be a label on YouTube, a credit line on Spotify, a filterable tag on Deezer and a royalty question on TIDAL. The only version that scales is one documented answer per track, delivered the same way everywhere.

Campaign architecture for this quarter

At MusicPromoToday, we have built campaigns since 2012, and every platform shift lands in the same place: whoever owns the plan owns the new field. Here is how disclosure plugs into each layer.

1. The distributor and label disclosure checklist

Before the next delivery, write one row per track:

• Composition: human, AI-assisted (which parts), or generated?
• Vocals: human performance, or synthetic or cloned voice?
• Instrumental layers: any generated stems, loops or sections?
• Lyrics: written, co-written with an AI tool, or generated?
• Visuals: music video, visualizer, Canvas-style loops and artwork. Any text-to-video or generated imagery?
• Production tools: pitch correction and AI mastering (No Gen AI on YouTube), logged so nobody over-declares out of nerves.

Map each row to the YouTube value (Fully, Partly, No), keep the same evidence for Apple and Spotify, and confirm your distributor actually passes the field through. When MusicPromoToday plans a release, this sheet sits next to the credits.

2. A PR narrative that matches the metadata

The fastest way to lose a story is a press line that contradicts the label a listener sees. If AI shaped the record, decide whether that is the story, a creative choice in the artist's own words, or a footnote in the credits. Either is defensible. "100% human" over a track declared "Partly" is not. For human-made records, process is now a story asset: the session, the players, the live take. Good music PR writes the disclosure answer into the media kit before a journalist asks.

3. Creator and UGC briefs that don't fake an artist

Creators are often where a listener meets a song first, so the brief needs a transparency clause. Never ask creators to present a generated vocal as a real singer or sell a backstory that isn't true. If AI tools touched the record or visuals, say so in the brief so creators aren't caught out in the comments. We covered the identity side in your name is now campaign risk, and the reach side in YouTube's Shorts originality update. Strong influencer and UGC marketing builds those clauses into the first brief.

4. Paid creative that doesn't look like an AI persona

Spotify has said AI-persona profiles will stay out of its recommendations by default, and listeners are learning the look. Ads built on faceless loops, generic synthetic visuals and an artist with no visible life pay to resemble the category platforms are labeling. Give paid real material: performance footage, studio clips, the artist's face and voice. Where AI sped up editing or versioning, keep the human artist at the center of the frame. Music advertising works best when every ad is clearly from someone.

5. YouTube Music and the main app as one surface

The label starts on the YouTube Music player and is coming to the main app, so official audio, video and Shorts will increasingly share one disclosure logic. Audit the whole channel: art tracks, music videos, lyric videos, visualizers. Make sure an AI visualizer isn't quietly turning a human record into "Partly" by accident. Then put proof of craft where fans can see it: live sessions, behind-the-scenes, performance cuts. A plan for YouTube music promotion should now include a disclosure pass alongside the thumbnail and title pass.

What to do this quarter

1. Run a catalog audit. Every upcoming track gets the checklist and a Fully, Partly or No answer.
2. Call your distributor. Ask how it delivers Gen AI declarations to YouTube, Apple and Spotify. Get it in writing.
3. Fix the video pipeline. Decide now whether AI visuals are worth a "Partly" designation on an otherwise human record.
4. Align the story. Media kit, bio, press release and creator brief should describe the process exactly as the metadata does.
5. Refresh paid creative. Lead with the artist, the performance and the room. Retire anything that reads as a synthetic persona.
6. Watch the next moves. YouTube's main-app rollout, Apple's "Made With AI" labels and Deezer's toggle are still coming. Re-check declarations when they land.

None of this asks you to stop using AI tools. It asks you to know where you used them and say so before a platform says it for you. MusicPromoToday has run 3,000+ campaigns across 40+ markets, and the releases that hold up are the ones where story, metadata and marketing match. If you want disclosure, PR, creators and paid built into one music marketing plan for your next drop, start with Plan My Release.

FAQ

What does YouTube Music's AI label mean?

Since October 14, 2026, YouTube Music shows a player-page label on songs created primarily using generative AI, such as tracks where a generative model created the composition and vocals. Tapping it shows more information. The labels will also expand to the main YouTube app.

How do I disclose AI use to YouTube?

Music partners declare Fully Gen AI, Partly Gen AI or No Gen AI through DDEX ERN 4.3.2+ or CSV delivery templates, or in the attribute section of YouTube Studio for direct uploads. Without a declaration, YouTube may use its own signals to designate content fully or partly Gen AI.

Will an AI label hurt a song's reach on YouTube?

YouTube has not said so. Neither its announcement nor its Help Center page says labeled tracks are handled differently in recommendations or royalties. Other services differ: Spotify has said it will keep AI Persona music out of recommendations by default, and TIDAL said it would stop paying royalties on fully AI-generated music.