Do you have to disclose AI generated video?
In most cases, yes. You have to disclose AI generated video when it is realistic enough that a viewer could reasonably believe it is a recording of something that actually happened. The major platforms all ask for this at upload, and several of them will attach a label themselves if their systems detect generated content in the file. Clearly stylized, animated, abstract or physically impossible material is generally outside the requirement. The safest working rule is simple: if it could be mistaken for real, label it.
The reason this question keeps coming up is that creators expect one rule and there are several. Each platform writes its own policy, uses its own wording, and updates it without much notice. That is why this post gives you a decision framework rather than a list of clauses to memorise. The framework outlives the policy pages.
What each platform asks for
The requirements differ in wording but converge on the same idea: realistic synthetic content gets identified, everything else largely does not. Below is the shape of it, described in terms of behaviour rather than exact policy text, because the text moves.
How the main platforms handle generated video at upload. Verify against the current policy before you rely on it.
| Platform | What it asks of you | What it may do on its own |
|---|---|---|
| YouTube | The upload flow asks whether your content is altered or synthetic and looks realistic | Adds a disclosure to the description, and a more prominent label on sensitive topics |
| TikTok | Asks creators to label realistic AI generated content | Reads content credentials in the file and can apply an AI generated label automatically |
| Instagram and Facebook | Ask creators to disclose realistic generated content | Apply AI information labels based on metadata and detection signals |
| LinkedIn and X | Policies are less prescriptive and change frequently | May surface content credentials where they exist in the file |
| Paid advertising | Ad policies are stricter than organic and vary by category and country | Rejects or restricts ads that misrepresent, especially in regulated categories |
Two things follow from that table. First, the metadata inside your file can trigger a label whether or not you ticked a box, so an undisclosed realistic video is not reliably an unlabelled one. Second, advertising is the strictest lane by a distance, which matters directly if you are producing the kind of work described in [how to make UGC style AI video ads](/blog/how-to-make-ugc-style-ai-video-ads).
The Disclosure Ladder
The Disclosure Ladder is the club's three rung test for deciding whether a given video needs a label. Work out which rung a piece sits on and the answer stops being a judgement call.
Three rungs, three different answers
| Rung | What it covers | What to do |
|---|---|---|
| Realistic depiction | A generated person, place or event that a viewer could take as a real recording | Always disclose, on every platform, in the upload settings |
| Assisted production | Real footage with generated b roll, AI upscaling, generated backgrounds, synthetic voiceover | Usually not required organically, but disclose in client work and any ad |
| Obviously synthetic | Stylized animation, surreal or impossible imagery, clearly artificial characters | No label needed, because nobody could mistake it for a recording |
The rung that trips people up is the middle one. A talking head that is genuinely you, cut with generated cutaways, is assisted production and generally does not need a label organically. The same video with a synthetic version of your face reading a script is a realistic depiction and does. The test is not how much AI was involved, it is whether the finished frame makes a claim about reality that is false.
Does an AI label reduce your reach?
Nobody can currently show you a trustworthy answer, and that is the honest position. The platforms do not publish reach data broken out by disclosure status, so there is no public dataset to reason from. Any specific figure you see quoted for how much an AI label costs in views is not verifiable, and it should be treated as noise rather than evidence.
What can be reasoned about is the mechanism. Distribution responds to watch time, retention and engagement. A label sits in the description or as a small marker, not in the first frames where retention is decided. So if a labelled video underperforms, the most likely explanation is the video, not the label. That is a hypothesis rather than a measurement, and it is described here as one deliberately.
The measurable risk runs the other way. Audiences react badly to feeling deceived, and a comment section deciding your content is fake does real damage to engagement, which is the thing distribution actually reads. Disclosure is cheap insurance against that outcome. If you want to test the effect on your own channel rather than take anyone's word for it, the metrics worth watching are in [the metrics that actually matter for AI video creators](/blog/the-metrics-that-actually-matter-for-ai-video-creators).
How to disclose without wrecking your hook
Disclose in the platform's own upload settings and nowhere near your opening seconds. The first three seconds decide whether the video is watched at all, and spending them on a compliance statement is the one genuinely costly way to disclose. Every method below satisfies the requirement without touching the hook.
- Use the platform toggle at upload. This is the version the platform recognises and the only one that counts as compliance.
- Put a plain line in the description, near the top, in ordinary language rather than legal phrasing.
- Add it to a pinned comment on platforms where descriptions are barely read.
- If you want it on screen, place it as a small persistent corner mark or an end card, never as a full frame opener.
- Keep the wording consistent across every post so it reads as a standard of yours rather than an admission.
Tone matters more than placement. Creators who disclose apologetically invite doubt about everything else they publish. Creators who state it flatly as a production fact are treated as professionals who know how their work was made. The second framing tends to become part of the brand rather than a cost to it.
One caution on burned in text: if you overlay a disclosure permanently into the frame, it travels with the file into every future use, including client deliverables and any [stock footage licensing](/blog/how-to-license-ai-video-as-stock-footage) you sell later. Keep disclosure in the metadata and the description where it belongs, and keep the master clean.
Disclosure in client work
In client work, disclosure is a contract question before it is a platform question. Decide in writing who is responsible for labelling at upload, because you usually are not the one pressing publish. A brand that does not know the footage was generated cannot comply, and the fallout lands on you regardless of who clicked.
- State in the proposal which shots are AI generated. Never let a client discover this from a comment section.
- Name who applies the label at upload, in writing, since the client normally controls the account.
- Ask early whether the brand has its own AI content policy. Larger organisations increasingly do, and it can rule out an approach before you build it.
- Flag regulated categories immediately. Health, finance, legal and political content carry rules well beyond platform policy and need proper advice, not a creator's judgement call.
- Never generate a likeness of a real person, employee or public figure without documented written permission.
Handled at proposal stage, this reads as diligence and it tends to build confidence rather than doubt. Handled after delivery, it reads as a disclosure you avoided. It also belongs alongside the other terms you should be setting before work starts, covered in [how to price AI video services for clients](/blog/how-to-price-ai-video-services-for-clients).
What happens if you do not disclose
The consequences escalate, and the first rung is often invisible to you. Platforms can apply a label themselves based on file metadata, so the practical result of skipping disclosure is frequently that you get labelled anyway, without control over the wording and with a compliance question attached to your account.
- A label applied for you, in the platform's wording rather than yours.
- Reduced distribution or removal where the content touches sensitive or newsworthy subject matter.
- Ad rejection, which is the fastest and most expensive failure mode in commercial work.
- Strikes or account level penalties for repeated undisclosed realistic content.
- Audience trust damage, which is the only one on this list you cannot appeal.
None of that is a reason to be nervous about publishing AI video. It is a reason to make disclosure a default step in your upload routine, the same way you already set a thumbnail and write a description. It takes seconds and it removes a whole category of risk from the operation.
Members inside Higgsfield Income Club compare how they word disclosures and what the current upload flows are actually asking for on each platform, which is the fastest way to keep up with rules that move. Join at higgsfieldincomeclub.com for $9/month and post the video you are unsure about before you publish it.
Short, practical drops on income paths, prompts that work, packaging for views, and getting paid. No spam, unsubscribe anytime.
Frequently asked questions
Do you have to disclose AI generated video on YouTube?
Yes, when it is realistic. The upload flow asks whether your content is altered or synthetic and looks realistic, and a disclosure is added where you say it is. Sensitive subject matter can carry a more prominent label. Clearly animated, stylized or impossible content is generally outside the requirement. Check the current upload flow, because the wording and the scope have both changed more than once.
Does TikTok automatically label AI generated video?
It can. TikTok asks creators to label realistic AI generated content, and it also reads content credentials embedded in the file, which means an AI generated label may appear whether or not you ticked the box. In practice that makes skipping disclosure a way to lose control of the wording rather than a way to avoid the label.
Do I need to disclose AI b roll in an otherwise real video?
Usually not for organic posting. Real footage with generated cutaways sits on the assisted production rung, where the finished frame is not making a false claim about reality. Disclose it anyway in client work and in any paid advertising, where the standard is stricter and the cost of a rejected ad is immediate.
Does an AI label hurt your views?
There is no trustworthy public data either way, because platforms do not publish reach broken out by disclosure status. Any specific percentage you see quoted is unverified. What is well established is that distribution follows retention and engagement, and a label does not sit in the opening seconds where retention is decided. Audience distrust, by contrast, does measurably damage engagement.
Who is responsible for disclosing AI video in client work, me or the client?
Whoever presses publish is responsible in practice, which is normally the client. That is exactly why it needs to be written down. State which shots are generated in the proposal, name who applies the label at upload, and confirm both in writing. A brand that does not know the footage was generated cannot comply, and the reputational damage still finds its way back to you.
Can I use a real person's likeness in an AI video if I disclose it?
Disclosure is not permission. Generating a recognisable likeness of a real person, whether an employee, a customer or a public figure, needs documented written consent, and in regulated or political contexts it needs proper legal advice as well. A label tells the audience how the video was made. It does not grant you any right to the face in it.
Last reviewed by David on August 22, 2026
