YouTube's AI content policy runs two separate tests. The first is disclosure: when you upload in YouTube Studio you must answer “Yes” to the AI use question if AI makes a real person appear to say or do something they did not, alters footage of a real event or place, or generates a realistic scene that never happened. The second is monetisation: the YouTube Partner Program judges AI-heavy channels on whether the work is original rather than mass-produced. YouTube says disclosing does not by itself cut reach or earnings, while repeatedly skipping a required disclosure can bring forced labels, removals or a YPP suspension.
This guide reads only YouTube's own pages: the Help Center article on disclosing GenAI content, the channel monetisation policies, the misinformation policies and two posts on the YouTube Official Blog. Eligibility thresholds are covered in our YouTube monetization requirements explainer; this page stays with the parts that AI changes.
Three rule sets that answer different questions
Most confusion about AI on YouTube comes from treating three systems as one. Each asks its own question, sits on its own page and carries its own penalty.
| Rule set | What it asks about your video | Where YouTube publishes it | Consequence of getting it wrong |
|---|---|---|---|
| The AI use disclosure | Could a viewer mistake generated or altered material for a real person, place or event? | Help Center: Disclosing use of GenAI content | A label applied by YouTube, and for creators who consistently skip it, removal or YPP suspension |
| Channel monetisation policies | Is the channel original and varied, made for viewers rather than purely for views? | Help Center: YouTube channel monetization policies | No YPP access, limited ads, or monetisation suspended on all or any of your accounts |
| Community Guidelines | Does it mislead in a way that risks serious harm, whatever label it carries? | Help Center: Misinformation policies, including manipulated content | Removal of the content, label or no label |
The answers are independent. A properly labelled realistic AI video can still fail a monetisation review because it is templated, and a highly original video can still breach the disclosure rule if a realistic synthetic scene goes unlabelled. YouTube's May 2026 post, Improving AI labels for viewers and creators, says a disclosure label alone does not change how a video is recommended or whether it can earn, and the Help Center makes the same promise about audience and earning eligibility.
What YouTube means by realistic
The controlling page is Disclosing use of GenAI content. It requires disclosure when AI is used to meaningfully alter or generate photorealistic content, and gives three triggers: making a real person appear to say or do something they did not do, altering footage of a real event or place, and generating a realistic scene that did not actually occur. Partial use counts; the page says content can be fully or partially altered or created with AI.
Two things fall outside the duty. Non-realistic AI content needs no disclosure, and neither do minor edits, which the page defines as changes that are mainly aesthetic and do not mislead viewers about what actually happened. The test is the viewer's likely belief about reality, not the tool you opened.
The scope has shifted since launch. YouTube's March 2024 announcement, How we're helping creators disclose altered or synthetic content, framed the tool around altered or synthetic media and said prominent player labels were for sensitive topics such as health, news, elections and finance. The current Help Center names the setting AI use, lists AI-generated music among the things you must disclose, and the 2026 post moves the label for all photorealistic content into a more visible position. If you learned the rules in 2024, re-check them.
Disclosure decision table, built from YouTube's examples
Each row is matched to an example YouTube publishes. Where YouTube gives no example, the row says so and offers a conservative reading. The lists on the Help Center are explicitly not exhaustive.
| What AI did in this upload | Select Yes under AI use? | Basis |
|---|---|---|
| Drafted the outline, script, title, thumbnail or an infographic | No | Help Center lists this as production assistance |
| Suggested video ideas or generated the captions | No | Help Center minor-use examples |
| Applied a beauty, colour, lighting, blur or vintage filter to your own footage | No | Help Center: aesthetic edits that do not mislead |
| Sharpened, upscaled or repaired old video, or cleaned up recorded audio | No | Help Center minor-use examples |
| Cloned your own voice to record a voice-over or a dub | No | Help Center names cloning one's own voice for voice-overs or dubs as minor |
| Generated a different real person's voice to narrate or comment | Yes | First trigger, and the 2024 announcement's example of synthetically generating a person's voice |
| Composed the soundtrack with an AI music generator | Yes | Help Center lists AI-generated music as requiring disclosure |
| Added invented footage of a real location to a promo or travel piece | Yes | Help Center example of extra AI footage of a real place |
| Showed a realistic weather disaster approaching a real city that never happened | Yes | Help Center example; also check the misinformation policy |
| Created a fantasy world, a fully animated sequence or an obviously impossible stunt | No, though YouTube may still show a note in the expanded description | Help Center non-realistic examples and label rules |
| Generated or extended a backdrop, such as scenery behind a parked car | No | Help Center minor-use example |
| Recorded gameplay from a video game | No | Help Center minor-use example |
| Ran a photorealistic AI presenter, not based on any real person, who talks to camera | Our reading: Yes | Not named by YouTube; a lifelike human speaking on camera is a realistic scene that did not occur |
The last row is the situation most AI-persona channels are in, and it is the one YouTube does not spell out. The reasoning for disclosing is simple arithmetic: an unnecessary disclosure costs you a label that YouTube says has no effect on reach or earnings, while a missed one can lead to a label applied for you and, if it becomes a pattern, penalties. YouTube's systems may also detect the AI and label the video regardless, as the next sections explain.
How to disclose in YouTube Studio, and what viewers see
The disclosure is a field in the upload flow, available in YouTube Studio on a computer or the mobile app. The Help Center gives these steps:
- Open YouTube Studio and start the upload as normal.
- Find the Attributes section and the question headed AI use.
- Select Yes if the video meets the disclosure requirements, or No if it does not.
- Carry on with the remaining video details and publish.
From late May 2026, according to the YouTube blog, the label for photorealistic and meaningfully altered or generated content sits directly below the player and above the description on long-form videos, and appears as an overlay on the video itself for Shorts. Unrealistic, animated or slightly altered content gets its note in the expanded description instead.
The two YouTube sources describe placement a little differently. The Help Center still says a player label “may also appear” for photorealistic content, while the blog calls the more visible label the single format for that content. The blog is the more recent statement, so the safe working assumption is that viewers of a realistic AI video will see the label without opening anything.
When YouTube applies the label for you
The Help Center says YouTube may label a video automatically when it was made with YouTube's own GenAI tools, when it contains C2PA metadata, or when internal systems detect AI generation or alteration. The blog adds that, from May 2026, these detection signals apply when a creator has not specified either way and the systems find significant photorealistic AI use. You get a notification in Studio when this happens.
If the label is wrong you can usually reverse it by answering No to the AI use question in the Attributes section. Some labels are locked. Here the sources differ in a way that matters:
- Help Center: labels on content made with YouTube's AI tools, content containing C2PA metadata, and content labelled after manual review cannot be adjusted.
- YouTube blog: disclosures stay permanent for content made with YouTube tools such as Veo or Dream Screen, and for content whose C2PA metadata shows it was fully generative AI.
The Help Center version is broader, because it does not limit the C2PA lock to fully generated files and it adds manual review. Plan around the stricter reading: if any clip in your edit came from a generator that writes provenance metadata into its files, expect a label you may not be able to remove, and decide before upload whether the edit still works with it.
How AI content is judged for monetisation
The YouTube channel monetization policies do not ban AI. They expect monetised work to be your own creation and not mass-produced, generic, repetitive or manipulative. Several passages now address AI directly:
- Template output. Under generic or repetitive content, AI-generated videos built from generic or unoriginal templates that give the impression of mass production, without your own insight or perspective, cannot monetise.
- Shock assembly. Under unsatisfying or off-putting content, the page rules out videos that stitch together unrelated or inconsistent AI clips to surprise viewers, and realistic visuals that trick people into believing a fake celebrity death or natural disaster happened.
- Assisted creativity. The same section gives allowed examples: using AI to visualise a character and narrative you invented, and using AI to edit your scripts or generate a background visual for a well-researched story.
- Synthetic experts. A channel using an AI-generated persona that presents itself as a human expert advising on health, legal issues, finances or politics will not be allowed to monetise.
Reviewers cannot watch everything, so the page says they may focus on your channel's main theme, most viewed and newest videos, the videos with the biggest share of watch time, your metadata and the About section. One high-performing AI compilation can therefore define how the whole channel is read. For a format-by-format originality check, use our faceless YouTube channel originality audit, and if a channel has already lost monetisation, the channel-level demonetisation fixes cover the review path. Separately, YouTube has confirmed Partner Program changes taking effect on 1 February 2027; our 2027 monetisation changes guide tracks who they affect.
What skipping the disclosure can cost
The Help Center says creators who consistently choose not to disclose may have a label applied manually or face penalties including content removal or suspension from the YouTube Partner Program. The enforcement section of the monetisation policies explains what suspension means in practice: monetisation can be suspended or permanently disabled on all or any of your accounts, and earnings linked to violations can be withheld or adjusted.
A label is also not a licence. The Help Center states that the Community Guidelines apply to all content, AI or not, and the misinformation policies prohibit technically manipulated content that misleads users and may pose a serious risk of egregious harm. One of YouTube's examples is video manipulated to make it appear a government official is dead. Disclosing that such a video is synthetic does not move it into the allowed column.
Upload checklist for AI-assisted videos
Run this before every upload that touched a generative tool. It takes a few minutes and leaves a record you can point to later.
- Write down every AI tool used in the edit, including voice, music, stock footage, background and thumbnail generators.
- For each one, ask the three trigger questions: does a real person seem to say or do something they did not, is real footage of a place or event altered, or does a realistic scene appear that never happened?
- Check the soundtrack separately, because AI-generated music is on YouTube's must-disclose list even when the visuals are untouched.
- If any answer is yes, or you cannot decide, select Yes under AI use in the Attributes section.
- Find out whether any clip came from YouTube's own AI tools or carries C2PA provenance data, and accept that its label may be permanent.
- If a real person's face or voice is simulated, hold their written permission before upload; the YouTube toggle records transparency, not consent. Our likeness licence checklist covers what that permission should say.
- If an AI host appears, confirm it does not give health, legal, financial or political advice as if it were a human expert.
- Compare the video with your previous uploads and ask whether a viewer would see a distinct idea, script and edit, or the same template with new prompts.
- Make sure the title, thumbnail and description do not present generated scenes as real footage.
- For dubbed versions in other languages, note which voice was used; see our guide to AI dubbing rules on YouTube and Meta before relying on a cloned voice.
- After publishing, check Studio for an automatic-label notification and, where YouTube allows it, correct a wrong label through the same AI use setting.
- Keep the tool list, your inputs and the disclosure answer with the project files. The record helps in a review and supports any later claim of human authorship, explained in our guide to copyright in AI-assisted content.
Limitations of this guide
This is general information drawn from YouTube's public pages as they read on 1 October 2026. YouTube changes these pages, its example lists are not exhaustive, and its reviewers apply judgement, so nothing here predicts how a particular video or channel will be treated. The photorealistic presenter row is our interpretation, not YouTube's wording. Laws outside YouTube can also require disclosure in places YouTube's toggle does not cover, including advertising rules for sponsored AI content, which our guide to AI influencer sponsorship disclosure sets out. If a monetisation decision affects your income, Partner Program members can raise it with YouTube Creator Support, and questions about consent or defamation in synthetic content belong with a qualified lawyer.