Yes, if the video looks real. As of September 2026, Meta, TikTok and YouTube all require an AI content label on realistic video or audio that was generated or significantly altered with AI. Clearly stylised animation and minor edits do not need one. Since 2 August 2026 EU law also requires deepfakes shown to people in the EU to be disclosed, while the UK and the US have no general labelling law and rely on the rules against misleading advertising.
We checked every rule below against the platforms’ and regulators’ own pages on 20 September 2026. These policies change often, so follow the links before you rely on them. This is general information, not legal advice.
What needs an AI content label on each platform?
Instagram and Facebook (Meta)
| Question | Meta’s rule |
|---|---|
| Needs a label | Photorealistic video or realistic-sounding audio that was digitally generated or altered, including with AI. Meta’s examples: a realistic video of people walking around a market, a reel narrated with a realistic AI voiceover, a song with AI vocals |
| Does not need one | A landscape video in a cartoon style. Still images: Meta does not require you to label them, although its systems may label them anyway |
| How to apply it | When creating a post or reel, switch on “Add AI label”. In Stories, open Options, then “Label AI” |
| Automatic labels | Meta reads industry-standard signals in the file (it names the C2PA and IPTC standards). Content created with AI gets a visible label. Content only modified with AI gets a note in the post’s options menu |
| If you skip it | “There may be penalties if you do not label content as required”, and Meta may apply the label itself |
Source: Instagram’s help page on labelling AI content. Instagram also has an optional “AI-generated profile” label, switched on under Edit profile.
TikTok
| Question | TikTok’s rule |
|---|---|
| Needs a label | All AI-generated content that contains realistic images, audio or video. TikTok also encourages a label on anything completely generated or significantly edited by AI |
| Does not need one | Minor corrections or enhancements. Content made only with TikTok’s own AI effects is labelled for you |
| How to apply it | Before posting, tap “More options” and switch on “AI-generated content”. A text note, hashtag sticker or a line in the description also counts as disclosure |
| Automatic labels | TikTok reads C2PA Content Credentials and may add an “AI-generated” label. You cannot remove an automatic label |
| If you skip it | TikTok may remove unlabelled realistic AI content under its Community Guidelines. Labelling real footage as AI is also a breach |
Source: TikTok’s support page on AI-generated content. Some content is banned even with a label: fake authoritative sources or crisis events, public figures shown making an endorsement, the likeness of anyone under 18, and the likeness of an adult private person without permission.
YouTube
| Question | YouTube’s rule |
|---|---|
| Needs a label | Realistic content that was meaningfully altered or generated: a real person appearing to say or do something they did not, altered footage of a real event or place, or a realistic scene that never happened |
| Does not need one | Clearly unrealistic or animated content, beauty filters, colour and lighting adjustment, sharpening and upscaling, and AI help with scripts or captions |
| How to apply it | During upload in YouTube Studio, answer “Yes” to the AI use question |
| Automatic labels | YouTube may add a label to content made with its own generative tools, content carrying C2PA metadata, or content its systems detect |
| If you skip it | Creators who consistently fail to disclose may get a label applied manually, have content removed, or be suspended from the YouTube Partner Program |
Source: YouTube’s help page on disclosing AI content. The label normally sits in the expanded description. For photorealistic content it may also appear on the video player.
Are the rules different for ads?
Yes, and this is where the three platforms differ most.
Meta. Ads built with Meta’s own generative AI tools are labelled automatically. The “AI info” note sits in the ad’s three-dot menu, or next to “Sponsored” when the ad contains an AI-generated photorealistic person. In a post updated on 1 June 2026, Meta said it will also begin detecting ads made with third-party AI tools through industry-standard signals. Its help page for advertisers still says third-party ads are not labelled automatically, so treat this as mid-rollout. We could not find, on Meta’s own pages, a disclosure toggle that ordinary commercial advertisers are required to use. Political and social issue ads have separate, stricter rules.
TikTok. Disclosure is mandatory. TikTok’s Business Help Center lists AI-generated, synthetic or manipulated media under mandatory disclaimers. In Ads Manager you tick “This ad contains AI-generated content” during ad creation. You cannot change it after submitting, and it resets whenever you create or duplicate a campaign, even if the ad was disclosed before. Spark Ads follow the organic labelling rules instead.
YouTube and Google Ads. In July 2026 Google began rolling out an AI label setting in Google Ads. The disclosure appears in the “How this ad was made” panel, and on the ad itself for campaigns targeting the EU, India and New York. When you upload a video to YouTube from Google Ads you are asked whether AI was used. Google may label assets itself, and says that using the setting does not guarantee compliance with any specific law. Election ads with synthetic content have their own disclosure requirement.
If you run AI ads on Meta, TikTok and YouTube, build these steps into trafficking. Our guide to AI UGC ads covers the rules on presenters and testimonials.
What does the law say?
EU: Article 50 of the AI Act
The transparency obligations in Article 50 have applied since 2 August 2026. According to the European Commission’s guidance:
- Whoever publishes a deepfake must disclose it clearly, at the latest when a person first sees it. A deepfake is AI-generated or manipulated image, audio or video that resembles existing persons, objects, places, entities or events and would falsely appear authentic.
- Evidently artistic, creative, satirical or fictional work still needs disclosure, but in a manner that suits the work.
- Providers of generative AI tools must mark outputs in a machine-readable way. Tools already on the market before 2 August 2026 have until 2 December 2026 to comply with that one duty.
- The rules reach companies outside the EU when the output is used in the EU. Fines can reach 15 million euros or 3% of worldwide turnover.
The Commission published a voluntary Code of Practice on 10 June 2026, with a set of EU icons for labelling. How far the deepfake definition reaches into ordinary advertising, for example an invented but realistic presenter, is not something we would call settled. We label that content anyway. The consent and advertising rules for a presenter like that are in our guide to AI spokesperson videos.
UK
There is no general statutory duty to label AI content. The ASA wrote on 29 May 2025 that “there is no blanket legal requirement in the UK to disclose the use of AI in ads”. Its test is whether the audience would be misled without disclosure, and it adds that disclosure alone will not save a misleading ad. In its March 2026 report on copyright and AI, the government said it proposes “to work with industry to explore best practice on labelling AI-generated content”.
US
There is no general federal labelling law. The FTC’s ordinary rules on deception apply, and states are moving on their own. New York’s synthetic performer law (General Business Law section 396-b), in effect since 9 June 2026, requires ads to conspicuously disclose an AI-generated human performer, with a civil penalty of 1,000 dollars for a first violation and 5,000 dollars after that. Audio-only ads and AI used only for translation are exempt. California’s AI Transparency Act became operative on 2 August 2026, but it binds large generative AI providers rather than brands posting video.
How to decide in practice
- A realistic generated person, place or event gets a label, on every platform.
- A realistic AI voice gets a label too. Meta names AI voiceover explicitly.
- Stylised animation and minor clean-up need no label.
- For TikTok ads, tick the AI box on every new and duplicated campaign.
- For Google and YouTube ads, answer the AI question honestly, and use the AI label setting when you target the EU, India or New York.
- Leave Content Credentials in your files. Stripping metadata to dodge a label is the wrong instinct.
- Never use a label to excuse a misleading claim. It does not work. Healthcare is the clearest case: our guide to what is allowed in reels for clinics and dentists explains why a generated before and after stays misleading with a label on it.
At Zero Shot, an AI video studio in London, we tell clients at delivery which files need the label on which platform, because an AI presenter and a stylised cartoon are treated differently. If you post yourself, our guide to making Instagram Reels with AI has a checklist to run before you post.
If you want AI video that arrives with the labelling worked out, send us a short brief.
Questions people ask
Does the AI label reduce reach?
TikTok says turning on its AI-generated content setting will not affect distribution, and YouTube says disclosure will not limit a video's audience or its eligibility to earn money. The Meta help pages we checked do not make a statement either way.
Do I need to label a video if AI only wrote the script?
No. The rules on all three platforms are about realistic generated or altered pictures and sound. YouTube lists script help, captions, beauty filters, colour adjustment and upscaling as uses that need no disclosure.
What happens if I do not label realistic AI video?
Meta says there may be penalties and may add the label itself. TikTok may remove unlabelled realistic AI content. YouTube may add a label, remove content or suspend repeat offenders from the YouTube Partner Program.
Is cartoon or stylised AI animation exempt?
From the mandatory label, yes, on all three platforms, because viewers would not mistake it for real footage. Platforms can still add an automatic label when the file carries AI metadata, and TikTok encourages labelling all fully generated content.