YouTube AI Policy: When You Must Label AI Content (and When You Don't)
Every time you upload to YouTube now, Studio asks a yes-or-no question in the Attributes section under "AI use." Most faceless creators use AI somewhere, for a script draft, a voice or a few shots, so the question looks like it applies to everyone.
It doesn't. YouTube's rule is narrower than the checkbox makes it look, and for a typical faceless channel the honest answer is often "No." This guide covers when the disclosure is required, how it applies to AI voices, images, b-roll and scripts, where the setting is, how the label looks to viewers, the automatic labels YouTube started applying in 2026, and what happens if you skip it.
The rule in one sentence
YouTube's help page puts it plainly: "We require creators to disclose AI-generated or altered content that seems realistic."
The key word is realistic. When YouTube introduced the disclosure in March 2024 (back then the tool was described as disclosing "altered or synthetic content"), it framed it around three kinds of content:
- using the likeness of a realistic person
- altering footage of real events or places
- generating realistic scenes
The current help page says the same thing in more detail. You need to disclose content that makes a real person appear to say or do something they didn't, alters footage of a real event or place, or generates a realistic scene that didn't happen. AI-generated music is now on the list too.
YouTube's own examples
The help page gives examples on both sides, and they're the best guide you'll get.
Content that needs disclosure includes:
- AI-generated music
- AI-generated extra footage of a real place, like a surfer in Maui for a travel promo
- realistic AI video of a match between two real professional tennis players
- making it look like someone gave advice they never gave
- a realistic tornado moving toward a real city that never happened
- making it look like a real person was arrested or imprisoned
Content that doesn't need disclosure includes:
- clearly unrealistic content, like someone riding a unicorn through a fantasy world
- an AI-generated animation of a missile in a fully animated video
- beauty filters, color and lighting adjustments, background blur
- using AI to create or improve an outline, script, thumbnail, title or infographic
- caption creation, idea generation, sharpening, upscaling, and voice or audio repair
- cloning your own voice for voiceovers or dubs
How it applies to a faceless channel
Here's how those examples map onto the parts of a faceless video. Where YouTube is explicit, we say so. Where it isn't, we tell you how we read it.
AI scripts. No disclosure needed. YouTube's 2024 announcement said it won't require disclosure when generative AI is used for productivity, "like generating scripts," and the help page lists scripts and outlines as production assistance. Our guide to writing YouTube hooks is still worth reading, though, because viewers can hear a generic AI script even when YouTube doesn't require a label.
AI voices. Cloning your own voice is listed as not needing disclosure. Making a real person appear to say something they didn't is listed as needing it, so a cloned voice of a real public figure reading words they never said needs the label. A generic synthetic narrator that isn't anyone in particular isn't named in either list. Our reading is that it doesn't meet the "real person" test, but YouTube hasn't said so directly, and its label placement rules do mention "non-music AI-generated audio." If your AI voice could be mistaken for a specific real person, disclose. See our guide to AI voices for faceless YouTube for picking one.
AI images. Thumbnails made with AI are production assistance, so no label. Inside the video, it comes down to realism. A painted-style illustration of a medieval battle is clearly not real. A photorealistic image of a real city flooding in an event that never happened is the kind of thing the tornado example describes.
AI b-roll. This is where faceless channels most often need the label. YouTube's own example is AI-generated extra footage of a real place. If your history video includes a realistic AI shot of a real street, building or event, disclose it. If the generated footage is stylized or animated, it falls under the "not realistic" side.
Stock footage and your own recording. If no AI made or changed the footage, there's nothing to disclose.
Where the setting is in YouTube Studio
According to the help page:
- Open YouTube Studio and start an upload.
- In the Attributes section, find "AI use."
- Select Yes if the content meets the requirements, or No if it doesn't.
- Continue with the rest of the video details.
YouTube says the setting is available in Studio on both a computer and a mobile device. Its May 2026 announcement also says that if you think a video was labeled by mistake, you can update its disclosure status in Studio.
How the label shows to viewers
YouTube changed how the label looks in May 2026. For photorealistic AI content, the label now appears directly below the video player, above the description, on long-form videos. On Shorts it shows as an overlay on the video. YouTube calls this the single label format for all photorealistic AI-generated or altered content.
For content that's unrealistic, animated or only slightly altered, the disclosure sits in the expanded description instead, in a section called "How this content was made." A separate help page says that section can show a "Made with AI" disclosure, and that content with C2PA Content Credentials can carry an "Info from" attribution naming who signed it.
YouTube now applies some labels automatically
This is the newer part, and it's in YouTube's own help page and its May 27, 2026 announcement. YouTube may add an AI label without you ticking anything for:
- content made with YouTube's own generative AI tools (the announcement names Veo and Dream Screen)
- content that carries C2PA metadata
- content YouTube's internal systems detect as AI-generated or altered
The announcement says automatic labels apply when the systems "detect significant photorealistic AI use" and the creator hasn't said whether they used AI. You'll get a notification in Studio when a label is applied automatically. If the systems get it wrong, you can usually change it by selecting No under AI use. Some labels can't be changed, though: those on content made with YouTube's AI tools, content containing C2PA metadata, and content labeled after a manual review.
Practically, this means skipping the question doesn't keep a label off a video that's full of realistic AI footage. It just means YouTube decides instead of you.
What happens if you don't disclose
YouTube says creators who consistently don't disclose may get a label applied manually, or face "penalties from YouTube, including removal of content or suspension from the YouTube Partner Program." Its Community Guidelines also apply to all content, AI-made or not.
The flip side is reassuring. YouTube says that disclosing AI content alone "won't limit a video's audience or impact its eligibility to earn money." The label isn't a penalty.
Just don't confuse the label with monetization approval. The disclosure rule is separate from the inauthentic content policy, which says repetitive or mass-produced content isn't monetizable whether or not it's labeled. A correctly labeled channel of templated AI videos can still fail a monetization review. Our YouTube monetization requirements guide covers what reviewers look at.
A simple rule for faceless creators
Ask one question about each video: could a viewer mistake any part of this for a real person, place or event that didn't actually look or happen that way? If yes, select Yes under AI use. If the AI only helped behind the scenes (script, outline, thumbnail, captions, your own cloned voice), select No.
When you're unsure, disclose. It costs nothing in reach or revenue, and it's a lot better than a manual label or a penalty later.
It helps to know where every shot in a video came from when you answer the question at upload. VidNexus keeps that record for you: stock clips from Pexels, Pixabay and Coverr show their source and licence, and an AI cinematic shot shows as one in the scene it belongs to. If a video uses a realistic AI shot of a real place or event, that's your cue to select Yes.