Labelling AI-Generated UGC on TikTok and Meta: What's Required
If your brand uses AI-generated UGC, you need to label it. Both TikTok and Meta now require disclosure of AI-generated or significantly AI-edited content, and both platforms can — and do — auto-label posts they detect as synthetic. Skipping the label is not a silent shortcut: it carries a real distribution cost.
This guide covers what the rules actually say, how to label correctly, and what to expect if you don't.
What the platforms require
TikTok asks creators to toggle "AI-generated content" when publishing any post that shows realistic-looking scenes, people or events created or significantly altered by AI. TikTok also runs its own detection and may automatically apply an AI-generated label. Failure to disclose is treated as a community-guidelines issue.
Meta (Instagram and Facebook) takes a similar approach: realistic AI-generated images, video and audio must be labelled by the uploader. Meta also auto-labels content where industry-standard AI provenance signals are detected.
The common thread: if it looks real and AI made it, disclose.
When labelling is required vs optional
Labelling is required when:
- A human face or voice is generated or cloned.
- A real scene is modified to show events that didn't happen.
- Audio is synthesised to sound like a specific person.
Labelling is generally not required for clearly stylised or animated content that nobody would mistake for footage of reality — illustrations, cartoons, obvious VFX.
When in doubt, label. There is no penalty for over-disclosing.
How to label correctly
- At upload, use the platform's built-in toggle. TikTok and Instagram both provide a setting in the publish flow. Use it instead of just writing "AI" in the caption.
- Be consistent across creators. If you commission AI UGC from creators, make labelling part of the brief. The creator publishing the post is the one responsible for the toggle.
- Document internally. Keep a record of which assets are AI-generated, which model produced them, and what human review took place. This is helpful for compliance, future audits and creative learning.
What happens if you don't label
- The platform may auto-label the content anyway, so the disclosure ends up on the post — but with a "platform detected" badge that reads worse than a self-applied one.
- Distribution may be reduced for content the platform considers undisclosed synthetic media.
- Repeated violations can affect the account's standing and, in serious cases, lead to removal.
The cost of disclosure is essentially zero. The cost of being caught not disclosing scales with how much volume you push.
What this means for performance
A common worry is that the AI label hurts watch time. The honest answer: not significantly when the content is well-made. Audiences already assume AI tools are involved in modern advertising. What hurts performance is low-quality AI output — uncanny avatars, robotic voices, generic B-roll — not the badge.
Treat labelling as a hygiene step, not a creative constraint. Spend the energy on making the content genuinely watchable.