AI disclosure scenario
Do you need to disclose an AI voice-over?
Direct answer
Often, but not in every workflow. TikTok requires labels for realistic AI-generated audio. Under EU Article 50, a voice-over requires disclosure when it constitutes a deep fake, such as a clone resembling an existing person; a generic synthetic narrator does not automatically meet that legal definition.
Check the identity, final audio, destination and audience. TikTok policy and EU deepfake duties are separate decisions.
Applies when
- The Article 50 deep-fake test covers AI-generated or manipulated audio that resembles an existing person, object, place, entity or event and would falsely appear authentic or truthful. A cloned executive or celebrity voice is therefore materially different from a generic narrator with no claimed identity.
- Where the duty applies, disclose by first exposure. Keep consent evidence separately for any identifiable voice: a label explains artificial origin but does not create permission to imitate a person.
Doesn't apply when
- Denoising, mastering or repairing a real recording without changing what the speaker said.
- On YouTube, cloning your own voice for voice-overs or dubs is listed as an example that creators do not need to disclose.
- AI-assisted script drafting followed by a real human voice recording.
Worked decision
Example
Real product footage uses a generic synthetic narrator for a TikTok campaign.
Decision: Disclosure required
The voice does not necessarily meet the EU deep-fake definition, but TikTok expressly requires a label for realistic AI-generated audio. The disclosure should name the narration, not the real footage.
Narration generated with AI. Footage is real.
Next steps
Required action
- Record the generated or altered component, source asset, tool and destination before publishing.
- Apply the native platform label wherever the applicable platform workflow requires it.
- Use component-specific wording such as: “Narration generated with AI. Footage is real.”
Publishing layer
Platform differences
TikTok
TikTok requires creators to label realistic AI-generated audio, including speech whose voice or words were altered by AI.
Open platform guide →YouTube
Disclose realistic generated or meaningfully altered content that makes a real person appear to say something they did not. YouTube lists cloning your own voice for a voice-over or dub among examples that do not require disclosure.
Open platform guide →Instagram and Facebook
Meta's AI info approach covers generated audio and video based on technical indicators or publisher disclosure; visible component wording remains useful where identity matters.
Open platform guide →Evidence
Source
Freshness and accountability
Last verified
Last reviewed 2026-09-23 · Current
Run this exact scenario
The content type, AI operation, context and starting platform for this scenario are preselected. Confirm the remaining facts.
Copy-ready language
Disclosure wording options
Narration generated with AI. Footage is real.
Voice generated with the speaker's documented consent.
Related guides
“Doesn't apply” means no disclosure requirement is identified for the facts stated here. A client policy, advertising rule or another destination platform can still be stricter. This page is source-backed information, not legal advice.