Labelling AI-generated video.
Reviewed
Short answer
An AI-generated or AI-manipulated video needs a clear disclosure in the EU when it is used professionally and is a deepfake: realistic footage resembling real people, places or events that could pass as authentic. The Code of Practice tells signatories to show the label at the start, at regular intervals where possible and at least after ad breaks, while AI backgrounds and standard special effects in film production usually do not make a video a deepfake.
Official sources: Commission guidelines on Article 50, section 6.1 · Code of Practice, Section 2, Measure 1.2
Deepfake or not: video examples from the guidelines
| Video | Result |
|---|---|
| An AI video of someone resembling a politician giving a speech | Deepfake: disclose |
| An AI-generated celebrity influencer in an advert | Deepfake: disclose |
| A realistic synthetic avatar of the company CEO congratulating employees | Deepfake: disclose |
| A teleshopping-style AI video with simulated customers using a product | Deepfake, and not an artistic work: full disclosure |
| A film or trailer with de-aged actors or digital replicas of dead actors, in a cinema or on a streaming service | Deepfake inside an artistic work: disclose in a way that does not hamper the film |
| Mice arguing about cheese in an advert | Not a deepfake |
| Real actors (not altered by AI) in front of an AI-generated prairie or ancient city | Not a deepfake |
| A science presenter with an AI animation of glaciers retreating | Not a deepfake |
The guidelines separate supporting effects from essential elements. AI-generated background scenes, special effects and technical pre- and post-processing in standard film production are unlikely to make a video falsely appear authentic. Fully AI-generated actors, digital replicas of real or deceased actors, de-aging, simulated performances, or untruthful scenes in a documentary are likely to.
Where the label goes in a video
- At the beginning of the video, and, where possible, at regular intervals, and at least after interruptions such as advertising breaks (Code of Practice, Section 2, Sub-measure 1.2.2(b), for signatories).
- Where people may join late (live streams, clips, screenshots), a disclosure only at the start is not enough; the guidelines ask for later disclosures where possible.
- In a closed, internal professional setting, such as a training video for staff, the Code allows the disclosure in the interface or setting before people see the video.
- Videos with sound are both image and audio: the provider must mark both, and an audible notice can support the visible one.
Platform rules for video
| Platform | Rule | Source |
|---|---|---|
| YouTube | Creators must disclose realistic AI content that makes a real person appear to say or do something they did not, alters footage of a real event or place, generates a realistic scene that did not happen, or creates music that is the main focus. Not required: unrealistic content, minor edits, production help such as AI scripts or thumbnails, cloning your own voice for voice-overs. Consistent non-disclosure can lead to removal or suspension from the Partner Program. | YouTube Help |
| TikTok | Creators must label AI-generated or significantly edited content that shows realistic-looking scenes or people; unlabelled content may be removed, restricted or labelled by TikTok. | TikTok |
| Facebook, Instagram, Threads | People must use the disclosure tool for organic posts with photorealistic video that was digitally created or altered; Meta may apply penalties if they do not. | Meta |
A platform’s “AI” toggle is the platform’s label. Under the AI Act the disclosure duty stays with you as the deployer; the Commission’s guidelines expect deployers in distribution chains to take proportionate measures, such as contract terms with distributors or interface settings, so the label is shown at first exposure.
Checklist before you publish an AI video
- Is anything in it realistic and resembling a real person, place, object or event? If not, no AI Act label is needed (platform rules may differ).
- Is the video evidently artistic, satirical or fictional? You still disclose, in a way that does not spoil it; the Code of Practice (Section 2, Commitment 3) names the description, the opening or end credits, or an icon next to the video frame.
- Label visible at the start and again after breaks; repeat during long or live videos.
- Tick the AI setting on each platform you upload to.
- Keep the source files and the generator’s Content Credentials.
Questions
- Does a fully animated AI video need a label?
- Not under the AI Act if nothing in it is realistic enough to pass as a real person, place or event: the guidelines give an advert with mice arguing over cheese as an example that is not a deepfake. YouTube likewise does not require disclosure for non-realistic content.
- Is ticking YouTube’s AI setting enough for the AI Act?
- It can be part of it, but the duty to disclose a deepfake at first exposure stays with you as the deployer. The Commission’s guidelines expect proportionate measures, including interface settings with distributing partners, so that the label is clearly shown to the audience.
Sources
- Guidelines on the transparency obligations under Article 50 of the AI Act (C(2026) 5054, 20 July 2026)European Commissionread
- Code of Practice on Transparency of AI-Generated Content (final text, PDF)European Commissionread
- AI Act, Article 50: Transparency obligationsEuropean Commission, AI Act Service Desk (AI Act Explorer)Consolidated text as at 27 July 2026 (includes the Digital Omnibus, Regulation (EU) 2026/1744), shown in the Commission’s AI Act Explorer.read
- Disclosing use of GenAI contentYouTube Helpread
- Community Guidelines: Integrity and Authenticity — Edited Media and AI-Generated Content (AIGC)TikTokread
- Labeling AI-Generated Images on Facebook, Instagram and ThreadsMeta Newsroomread
Last reviewed by Volkan Çatak, Founder, Proofcan.
Not legal advice. This page explains official texts in plain words and links each one; check the sources, or a lawyer, before relying on it for a decision.