01
The rule everywhere: realistic
YouTube, TikTok and Meta all draw the line at realism. If a viewer could mistake AI-made content for something real, a person saying words they never said, a real event changed, a scene that never happened, you must label it. Using AI to write a script, fix colors or blur a background doesn't count.
Where they differ is sound. Meta requires a label for realistic-sounding AI audio, including a human-sounding AI narrator. YouTube and TikTok don't, unless the voice imitates a real person: TikTok explicitly exempts generic text-to-speech narration. YouTube does require a label when AI-generated music is the focus of the video. Clearly unrealistic styles, such as anime, need no label on any of them.
02
YouTube: altered or synthetic content
YouTube requires disclosure when realistic content makes a real person appear to say or do something they didn't, alters footage of a real event or place, generates a realistic scene that didn't occur, or creates music that is the main focus of the video. You set it in YouTube Studio during upload; YouTube adds a label to the description, and for photorealistic content possibly to the player.
YouTube says disclosure doesn't limit a video's audience or monetization. Repeatedly skipping it can lead to a label added by YouTube, removal or suspension from the Partner Program.
03
TikTok and Meta
TikTok's Community Guidelines require a label when AI or significant edits show realistic-looking people or scenes: a face swapped, someone made to say something they didn't, an object or person added or removed misleadingly, or AI audio that mimics a real person's voice. Small edits, artistic styles and generic text-to-speech narration need no label. Use the AI-generated content label, or a clear caption, sticker or watermark.
Meta requires its disclosure tool for organic posts with photorealistic video or realistic-sounding audio that was digitally created or altered, and says it may apply penalties when people don't. The visible label is “AI info”; images can also be labeled automatically from industry metadata.
04
The EU AI Act from August 2, 2026
Article 50 of the EU AI Act applies from August 2, 2026. Anyone who publishes a deepfake in a professional capacity, AI content that resembles existing people, objects, places or events and would falsely appear authentic, must disclose that it is artificially generated or manipulated. It doesn't cover purely personal use, and for evidently artistic, satirical or fictional work the disclosure can be lighter.
Unlike the platform rules, this is law, and it applies whichever platform you post on. A platform label usually satisfies it in practice, but a clear mention in the video or caption is the safer habit.
Sources and methodology
Numbers checked on October 8, 2026.
- Disclosing use of altered or synthetic content — YouTube Help, October 8, 2026
- Community Guidelines — Integrity and authenticity (edited media and AIGC) — TikTok, October 17, 2026
- Labeling AI-Generated Images on Facebook, Instagram and Threads — Meta Newsroom, April 1, 2025
- Our Approach to Labeling AI-Generated Content and Manipulated Media — Meta Newsroom, September 12, 2024
- Regulation (EU) 2024/1689 (AI Act), Article 50 — EUR-Lex, July 12, 2024
