# AI content labels: platform rules and creator disclosures

Prompt and Render desk. A ZeroClick publication (https://promptandrender.com/about/). Published 2026-10-15.
Format: Explainer. Topic: Rights and rules (https://promptandrender.com/topics/rights-and-rules/).
Canonical: https://promptandrender.com/articles/ai-content-labels-platform-rules-and-creator-disclosures/

An automatic AI label is a platform's reading of your file; a disclosure is your statement about what you made. YouTube, for instance, can apply a label itself but still [requires creators to declare specified AI content](https://support.google.com/youtube/answer/14328491?hl=en). A missing badge is no reason to skip the upload question.

## Why can an unlabeled file still need disclosure?

The trigger for an automatic label and the trigger for your declaration need not be the same. YouTube names its own GenAI tools, C2PA metadata and internal detection as automatic triggers. Its upload rule instead names realistic inventions or meaningful alterations, and AI music that is the video's main focus.

Consider a photorealistic clip of an imaginary fishing harbour, made for a travel promo. Removing its metadata would not make that scene a real recording. YouTube's realistic-scene rule concerns what was generated, not whether a marker survived the export.

The two routes can also end in the same visible badge. Meta's [AI info policy](https://about.fb.com/news/2024/04/metas-approach-to-labeling-ai-generated-content-and-manipulated-media/) says the label can follow industry signals or the uploader's disclosure, and viewers can see which prompted it. The badge alone does not tell you that a detector made the decision.

## What does each platform ask for?

The table separates the platform's signal from your publishing action. Policies checked October 14, 2026; the Meta ads row covers its published labeling rollout, rather than every advertising requirement.

| Platform / scope | Automatic trigger | What you submit | Stated consequence of omission |
|---|---|---|---|
| YouTube Studio uploads | Own GenAI tools; C2PA; internal systems | AI use: Yes or No, under Attributes | Consistent non-disclosure: manual label, removal or Partner Program suspension |
| [Meta organic posts](https://about.fb.com/news/2024/02/labeling-ai-generated-images-on-facebook-instagram-and-threads/) | C2PA/IPTC indicators | Disclosure tool for digitally created/altered photorealistic video or realistic audio | Penalties possible; not specified |
| [Meta ad-labeling rollout](https://about.fb.com/news/2025/02/gen-ai-transparency-metas-ads-products/) | Own ad tools; third-party signals rolling out | Page describes automatic labeling only | Omission penalty not addressed there |
| [TikTok posts](https://support.tiktok.com/en/using-tiktok/creating-videos/ai-generated-content) | AI effects; C2PA Content Credentials | AI-generated content setting, text, hashtag sticker or description | Unlabeled AI content may be removed |
| [Spotify song credits](https://support.spotify.com/us/artists/article/ai-credits/) | None described on credits page | Optional role-level credits through supported distributors | No penalty given on credits page |

## Where do the video platforms draw their lines?

YouTube exempts non-realistic work and minor edits, including colour filters, repair and upscaling. Production help with a script or thumbnail also needs no declaration. Most erroneous automatic labels can be corrected in Studio, but labels tied to YouTube's AI tools, C2PA or manual review cannot be adjusted.

For music delivered through a distributor, the declaration may travel with the release instead of an ordinary upload. YouTube's [music-partner guidance](https://support.google.com/youtube/answer/17124251) describes Fully, Partly or No Gen AI declarations through DDEX or CSV. An omitted value means unknown unless other signals let YouTube decide. That is a delivery route to check with the distributor, not a reason to leave the Studio question unanswered.

Meta's organic rule covers Facebook, Instagram and Threads. It requires disclosure for digitally created or altered photorealistic video and realistic-sounding audio. That is narrower than a demand to declare every retouched still. Meta's September 2024 policy update also distinguishes generation from editing: edited-only content gets AI info in the post menu, while generated content retains the displayed label.

Marketers buying ads need to check a different surface. Meta's June 1, 2026 update describes About this ad, a menu destination for AI info labels, and a rollout of labeling from third-party AI signals as well as Meta's own ad tools. The original February 2025 text put some labels beside Sponsored; the June update describes the newer menu destination and notes regional variation. Its account of automatic labels cannot settle every advertiser's disclosure obligations.

TikTok encourages labels broadly for work wholly generated or substantially changed by AI, but requires them for realistic AI images, sound or video. Its help page describes Content Credentials as metadata attached to a file. Those credentials and TikTok's AI effects can trigger automatic labels that the poster cannot remove.

There is a limited shortcut for work made only with TikTok's AI effects: it may already be labeled, without another creator label. Extra AI changes made outside the effect still come under the disclosure guidelines. And a label cannot make prohibited content acceptable: TikTok says some AI content remains forbidden even when labeled.

## Why are Spotify AI credits different?

Spotify's credits identify an AI contribution to a role, such as lyrics, vocals or production. They do not brand the whole track as AI-generated. Credits remain optional, in beta, visible on mobile in song credits and Now Playing, and available through participating distributors.

For a song with generated instrumentals and human vocals, the useful distinction is which contribution belongs to which role. Spotify asks for separate credits when both AI and humans contributed to the same role. A creator can describe the music more precisely than a single yes-or-no badge would allow.

The separate [AI Persona badge](https://support.spotify.com/us/artists/article/ai-personas/) concerns an artist's public identity as an AI-generated photorealistic human. It can follow self-disclosure or Spotify's review; the latter appears as Likely AI Persona. Neither route is a judgment that the song itself was generated.

## What does Article 50 require in the EU?

The [EU AI Act](https://eur-lex.europa.eu/legal-content/EN/TXT/PDF/?uri=OJ%3AL_202401689) separates the company supplying an AI system from the person or organisation using it. Article 50 gives providers a machine-readable marking duty for synthetic output, with exceptions for standard editing assistance and changes that do not substantially alter input data or its meaning.

For images, audio and video, the deployer's disclosure duty concerns deepfakes. The Act defines these by resemblance to existing people, objects, places, entities or events, combined with a false appearance of authenticity or truth. Evidently creative, fictional or satirical work still calls for disclosure, but in a form that preserves the experience of the work. Information must be clear and distinguishable by first exposure.

Article 2 sets the Act's scope, including EU deployers and some outside the EU whose output is used there. It excludes deployer obligations for natural persons acting purely personally and non-professionally. A platform checkbox does not decide those scope questions.

Article 50's general application date was August 2, 2026. The [July 2026 amendment](https://eur-lex.europa.eu/legal-content/EN/TXT/PDF/?uri=CELEX%3A32026R1744) gives providers of systems marketed before that date until December 2, 2026 to meet the output-marking duty. It does not postpone deployers' deepfake disclosure duty. These are separate responsibilities with separate timing.

## What should you keep before publishing?

Our recommendation is to keep the source asset and a short record of what was generated, what was edited and with which tool. Preserve the provenance carried by the export, then answer the destination's declaration separately. That gives you both a file history and a record of what you told the platform.

A travel clip, a retouched photograph and a song with generated instrumentals do not arrive at the same upload question. Check the work against the rule for its destination, rather than waiting for the destination to recognize the work.

## Frequently asked questions

### Does removing AI metadata remove the need to disclose?

No. Removing metadata changes a signal, not what you made. Our reading is that the stated disclosure duties still apply, including Meta's rule for digitally created or altered photorealistic video and realistic audio.

### Can I remove an automatic TikTok AI label?

No. TikTok says the poster cannot remove its automatic AI-generated label, including labels prompted by AI effects or Content Credentials.

### Are Spotify AI credits mandatory?

No. Song credits are optional, role-specific and submitted through participating distributors; AI Persona badges concern public identity separately.

### Does Article 50 require disclosure of every AI-assisted image?

No. For images, audio and video, its deployer disclosure duty concerns deepfakes within the Act's scope, while providers have a separate machine-readable marking duty with stated exceptions.