YouTube Music on October 14, 2026 began rolling out labels that tell listeners when a song was created primarily with generative AI, according to a post on the YouTube Official Blog. The label sits on the player page, opens a menu with more detail when tapped, and is applied using a mix of information from music partners and YouTube's own detection tools. The company said the labels will also expand to the main YouTube app, without giving a date.

In Short

YouTube Music now puts a small tag on songs that were mostly made by an AI tool, for example when a computer wrote both the tune and the singing. This matters for listeners who want to know whether a person made what they are hearing, and for artists competing with a growing flood of machine-made tracks. You will see the tag on the screen where the song plays, and YouTube says the same tags are coming to the regular YouTube app later.

What YouTube published

The announcement is short. It runs to four paragraphs on the YouTube Official Blog, filed under the "YouTube Music" and "AI" topics and dated October 14, 2026. No executive is named, no figures are given and no markets are listed.

The post opens by framing the change as a response to listeners. "Generative AI is opening up new ways for artists to create, experiment, and share music," according to YouTube. "As these tools evolve, our listeners have told us they want transparency when they're hearing an AI-generated track."

The operative sentence follows: "Starting today, YouTube Music is rolling out labels to make it easier to see when a song was created primarily using generative AI." YouTube gave one example of what crosses the line, stating that "this includes using a generative model to create the composition and vocals."

That example matters, because it sets a high bar. A track in which a human writes and sings while an AI tool handles mixing, mastering or a backing loop would not obviously qualify under the example given. A track in which a model produced both the melody and the voice would.

Where the label appears

According to the post, the label is placed on the player page, the screen that shows the currently playing track. "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," YouTube wrote. "Tapping the label opens a simple menu that shares more information."

The post does not reproduce the label's wording, describe what the menu contains, or say whether the label also appears in search results, playlists, artist pages or the queue. Spotify, by contrast, listed four surfaces for its own AI badge when it described it in August, a comparison taken up further below.

Two thresholds in four paragraphs

Readers comparing the paragraphs will notice two different formulations. The opening paragraph says labels mark songs "created primarily using generative AI." The second paragraph says they appear on tracks "where AI played a significant role."

"Primarily" and "significant" are not the same test. A significant role can fall well short of a primary one. YouTube did not say which standard its systems apply, whether the two phrases are meant as synonyms, or where a track with an AI-generated vocal over a human-written composition would land. The post refers readers to the YouTube Help Center for "more details on disclosure guidelines and standards," but the blog itself does not resolve the difference.

How tracks get identified

YouTube described a two-source approach. "We're teaming up with music partners and using our own detection tools to identify AI-generated tracks, applying the same labeling principles already used across YouTube," the post states.

Neither source is detailed. The music partners are not named, and it is not stated whether they are record labels, distributors or rights societies, or whether they supply information through delivery metadata, contractual declarations or another channel. The detection tools are not described either. There is no accuracy rate, no false-positive rate, no mention of whether a human reviews flagged tracks, and no word on whether an artist or label can contest a label once applied.

The reference to "the same labeling principles already used across YouTube" points to a system PPC Land has followed for more than two years. YouTube first required creators to disclose realistic altered or synthetic content in March 2024, with labels placed in the expanded description for most videos and on the player for sensitive subjects such as health, elections and finance. That requirement took formal effect on May 21, 2025, when YouTube made the "altered content" setting in YouTube Studio mandatory for realistic content and warned that repeated non-disclosure could lead to removal or suspension from the YouTube Partner Program.

The most recent change came in May 2026. YouTube moved its generative AI labels to more visible positions: an overlay on Shorts, and a line directly below the player on long-form videos, instead of inside the collapsed description. That update also set out three routes to an automatic label when a creator does not answer the AI question - content made with YouTube's own generative tools, content carrying C2PA provenance metadata, and content flagged by internal classifiers.

Whether YouTube Music inherits those three routes is not stated. If it does, the "music partners" in the post may correspond to declared metadata, and the "detection tools" to the internal classifiers. That reading is inferred from the earlier policy, not stated by YouTube.

Google's own music generators

Google is not a neutral party here. It builds the generators as well as the labels. Lyria 3.5 reached the Gemini app and the Gemini API on September 4, 2026, producing songs of a couple of minutes with verses, choruses and bridges at 44.1 kHz stereo. Every Lyria track carries a SynthID watermark, an imperceptible signal embedded in the audio waveform rather than in a metadata field that can be stripped.

YouTube has its own history with these tools. Dream Track ran as a limited experiment from late 2023 with nine recording artists, and YouTube integrated Lyria 2 through a Speech to Song feature in September 2025, according to the same PPC Land report. A watermark embedded by Google's own model gives YouTube an obvious detection signal for Google-made audio. Music produced on competing generators carries no such guarantee, which is where classifiers and partner declarations would have to do the work.

The October 14 post does not mention SynthID, C2PA or Lyria by name.

The rest of the streaming market

YouTube is not first. Spotify set out its approach on September 25, 2025, combining AI credits delivered through the DDEX metadata standard, an AI-aware spam filter and a ban on unauthorised voice clones. Fifteen labels and distributors committed to the DDEX standard at the time, PPC Land reported in June 2026.

Spotify went further in August. On August 13, 2026, it introduced an AI Persona badge for artist profiles whose public identity appears machine-generated, with badges visible to listeners from mid-September and shown on profile banners, About sections, search and track rows in playlists. Badged profiles are excluded by default from editorial and algorithmic recommendations. According to Spotify, the badge concerns an artist's public identity rather than how the music was made - a different target from YouTube's, which labels individual songs according to how they were produced.

That difference is worth stating plainly. YouTube labels tracks; Spotify's newer badge labels artists. And YouTube's post says nothing about recommendations. Spotify demotes badged profiles. Deezer removes AI-detected songs from algorithmic recommendations and editorial playlists. YouTube's announcement describes a label and a menu, nothing more. Whether a labelled track is treated any differently by YouTube Music's recommendation systems, or by royalty accounting, is unaddressed. Under YouTube's broader policy, a label by itself does not change recommendations or monetisation eligibility, according to the May 2026 documentation PPC Land reviewed.

Deezer's numbers

The scale of the problem is clearest in Deezer's disclosures. The French service receives about 75,000 AI-generated tracks a day, more than 44% of its total daily deliveries, up from 30,000, or 28%, in September 2025, according to figures it published in June 2026. Fully AI-generated music accounted for 1% to 3% of streams on Deezer, and up to 85% of streams on those tracks in 2025 came from stream manipulation, the company said.

Deezer also commissioned an Ipsos survey of 9,000 people in eight countries. It found that 80% wanted fully AI-generated music clearly labelled, and that 97% could not correctly identify the human-made track in a blind test. Those findings come from a company selling detection technology - Deezer licensed its detector to the French collecting society Sacem in January 2026 - and were not independently audited. Even so, they track closely with the reason YouTube itself gave: listeners "want transparency."

YouTube published no comparable figures. It did not say how many AI tracks it receives, what share of its catalogue it expects to label, or how many songs carried labels at launch.

Licensing, not only labelling

Labels sit alongside a separate shift in how AI music is licensed. Universal Music Group and Udio settled their copyright litigation on October 29, 2025 and agreed to launch a licensed creation platform in 2026, with artists setting permissions. The article listed YouTube among UMG's earlier AI agreements, alongside TikTok and Meta. Artist groups have pressed for more. The Music Artists Coalition set out three principles on October 30, 2025 - consent, compensation and transparency - and questioned whether settlement money would reach artists.

A label tells a listener how a song was made. It does not say whether the model behind it was trained on licensed recordings, or whether anyone was paid.

The regulatory backdrop

The timing coincides with European rules taking effect. Article 50 of the EU AI Act, which covers transparency for AI-generated content, became legally applicable on August 2, 2026. The European Commission published guidelines and a voluntary Code of Practice on July 20, 2026, requiring providers of systems that generate synthetic audio, images, video or text to mark outputs in a machine-readable format. Fines reach EUR 15 million or 3% of worldwide annual turnover, whichever is higher.

Google announced on July 24, 2026 that it would sign the code. The code calls for multi-layered marking of audio, combining signed metadata with an imperceptible watermark, and sets two further dates: December 2, 2026, for generative systems already on the market to bring their marking into conformity, and February 2, 2027, for signatories to implement an interoperable watermark-detection solution.

Those obligations fall mainly on the providers of generators and on deployers who publish deepfake content. A streaming service displaying a label to listeners is a different role. YouTube's post does not cite the AI Act, and nothing in it ties the rollout to European law. Still, a cross-industry detection standard due by February 2027 would give platforms such as YouTube Music more machine-readable signals to work with than they have at present.

Why it matters for marketers

For advertisers, the most direct question is where labelled music appears next to paid media. YouTube Music's free tier carries ads; its Premium tier does not. Premium and Music Premium had more than 125 million subscribers including trials as of March 2025, and YouTube's upgraded Ask Music assistant - covering more than 300 million songs - is available only to those subscribers, PPC Land reported in September. The October 14 post does not say whether labels are visible in both tiers, though nothing in it suggests otherwise.

The expansion to the main YouTube app is the larger matter. Music on YouTube is not confined to music videos; it runs under Shorts, creator uploads and background tracks across the platform. A song-level AI label in the main app would add a second disclosure layer beside the existing creator-declared "altered content" label, and the post does not explain how the two will interact.

Brand safety vendors already treat AI-generated content on YouTube as a category. Integral Ad Science extended its Low-Quality GenAI Avoidance product to YouTube on September 30, 2026, using YouTube Data API signals that, according to IAS, Google does not review. Earlier, a Kapwing study found that 104 of the first 500 Shorts shown to a new account were AI-generated slop, with top channels estimated to earn $4 million to $4.25 million a year from ads. A first-party YouTube label on AI music is a signal that verification companies may eventually want to read, though whether YouTube will expose it through any API is not stated.

Advertisers are also on the other side of this rule. Since July 2026, Google has rolled out an AI labelling setting across Google Ads, Display & Video 360, Campaign Manager 360, Merchant Center and Ads Editor, with visible overlays required for campaigns targeting the EU, India and New York. An advertiser using a Lyria-generated jingle in a YouTube ad already faces Google's ad-side disclosure regime. Whether YouTube Music's song labels and Google's ad labels draw on the same detection systems is another point the post leaves open.

What remains unanswered

The gaps in a four-paragraph post are many. Which markets get the labels first? Which partners supply data? What is the threshold - primary or significant? Can artists dispute a label? Does a label affect recommendations, editorial playlists or payouts? What does the menu say? When does the main YouTube app follow?

YouTube's answer to most of these, for now, is a pointer to the Help Center. The labels began rolling out on October 14, 2026. How far and how accurately they reach remains to be seen.

Timeline

Summary

Who: YouTube, through its YouTube Music service, working with unnamed music partners and its own detection tools. Listeners, artists, labels and advertisers on YouTube are affected.

What: YouTube Music began rolling out labels identifying songs created primarily with generative AI, such as tracks where a model generated both the composition and the vocals. The label appears on the player page and opens a menu with more information. The post uses two thresholds - "primarily" and "significant role" - and gives no markets, partners, accuracy figures, appeal process or effect on recommendations.

When: The rollout started on October 14, 2026, the date of the YouTube Official Blog post. Expansion to the main YouTube app was announced without a date.

Where: In the YouTube Music app, on the player page, with the main YouTube app to follow. No countries were specified.

Why: YouTube said listeners told it they want transparency when hearing AI-generated tracks. The move follows Spotify's DDEX AI credits and AI Persona badge, Deezer's report that more than 44% of its daily deliveries are AI-generated, and the EU AI Act's Article 50 transparency obligations, which became applicable on August 2, 2026.