Since AI has become more prevalent (and much more realistic), YouTube requires creators to disclose when a video contains realistic AI-generated or AI-altered content that a viewer could mistake for real footage. The rule targets deception, so writing a script with AI, generating a thumbnail, or using a generic text-to-speech voice doesn’t require a label.
Realistic synthetic content that could pass as real footage needs the "altered or synthetic content" disclosure. Everyday production assistance is exempt. Skip it when required, and YouTube can apply the label for you.
|
Question |
Answer |
|
What is it? |
A transparency label for realistic AI-generated or AI-altered video content |
|
When is it required? |
When content could be mistaken for real footage of a real person, place, or event |
|
Where does the label appear? |
In the video description; for photorealistic and sensitive-topic content, also directly on the video player |
|
What if you skip it? |
YouTube can force the label on; consistent non-disclosure risks content removal or YPP suspension |
What Is YouTube's Altered or Synthetic Content Disclosure?
The altered or synthetic content disclosure is a setting in YouTube Studio that creators enable when a video contains realistic AI-generated or AI-altered media a viewer could mistake for real-life footage. YouTube announced the framework on November 14, 2023, and the Studio toggle went live in March 2024. Disclosure doesn’t make otherwise-prohibited content available, because Community Guidelines still apply whether or not something is labeled.
What Triggers the Disclosure Requirement and What Doesn't
|
Requires disclosure |
Example |
|
Realistic synthetic face or voice |
An AI-generated person delivering what looks like a real news broadcast |
|
Cloned voice of a real person |
Audio altered to make a public figure appear to say something they didn't |
|
Fabricated real event |
AI-generated footage that appears to show a real disaster, protest, or accident |
|
Altered footage of a real place |
Synthetic scenery making a real location look different than it is |
|
Deepfake of a real person |
Video making a real individual appear to say or do something they didn't |
Okay, we learned what requires AI disclosure, but what doesn’t?
|
No disclosure needed |
Example |
|
AI-assisted scripts, titles, outlines |
Using an AI tool to write copy |
|
AI-generated thumbnails |
Still images made with an AI image tool |
|
Generic AI voiceover |
Text-to-speech narration in a clearly synthetic voice, not impersonating anyone |
|
Clearly animated or stylized visuals |
Cartoons, motion graphics, 3D animation |
|
Unrealistic special effects |
Fantasy explosions, sci-fi environments |
|
Standard edits |
Filters, color correction, background blur |
|
AI-generated captions or translations |
Text-layer tools |
The difference is in realism: disclosure applies when a reasonable viewer could believe the content depicts something real. YouTube’s guidance puts it like this: creators aren’t required to ‘disclose content that is clearly unrealistic, animated, includes special effects, or has used generative AI for production assistance.’ More on AI policies on YouTube here.
Where the Label Appears on YouTube
Photorealistic AI content that meets the disclosure threshold gets a label directly on the video player, and YouTube applies this placement more consistently on sensitive subjects like elections, ongoing conflicts, and public health.
Content that's AI-assisted but not photorealistic (stylized or animated AI content, for example) typically only gets the label in the expanded video description, under a heading identifying it as altered or synthetic content.
How to Add the Disclosure in YouTube Studio
- Upload or edit a video → Details → Attributes section
- Find "Altered or synthetic content" and toggle it to Yes
- Save
The disclosure can also be added after a video is already published:
YouTube Studio → Content → select the video → Details → update the same Attributes toggle
What Happens If You Don't Disclose
YouTube can identify undisclosed synthetic content itself, through its own internal detection systems, C2PA metadata embedded in a file, or manual review, and apply the label without the creator's input.
Since May 2026, that automated detection has expanded to specifically detect photorealistic AI content, applying labels directly on Shorts and beneath long-form videos, even when a creator never toggled the setting. A label applied this way generally can't be removed by the creator, and it's a permanent restriction for videos made with YouTube's own AI tools or ones carrying C2PA metadata that fully identifies them as AI-generated.
Beyond the forced label, YouTube has stated it will look at further enforcement for creators who consistently choose not to disclose, up to content removal or suspension from the YouTube Partner Program for repeat cases.
A note on the policy timeline:
Mandatory enforcement for this disclosure requirement went into effect on May 21, 2025, when YouTube's systems began actively scanning uploads for undisclosed synthetic content. A separate policy — renamed from "repetitious content" to "inauthentic content" on July 15, 2025, and tightened further in a January 2026 enforcement wave — targets mass-produced and templated content more broadly, AI or not, and affects monetization eligibility rather than disclosure. The two shouldn't be confused.
Does Disclosing Affect Reach or Monetization?
No. YouTube has said directly that adding the disclosure label doesn't affect a video's recommendations, search ranking, reach, or monetization eligibility. The penalty structure applies to not disclosing when required.
This article explains how YouTube's disclosure system works; it isn't legal advice on AI content regulation.
