Calabi Labs · Guide · 2026-05-27
Ai disclosure rules by platform youtube instagramfacebook and tiktok l
Now I have comprehensive, current data across all four platforms. Here's the full answer:
AI Disclosure Rules by Platform: YouTube, Instagram/Facebook, and TikTok
Here's a plain-English breakdown of what each major platform requires in 2025, what triggers a label, and how to stay compliant.
YouTube
Effective: March 2024 (announced) — enforcement ramped up in early 2025.
What triggers a label: YouTube requires creators to disclose any realistic synthetic or AI-generated content — video, audio, or imagery — that could be mistaken for real footage. This includes AI-generated faces, voices, or events that never happened.
How to label: During upload, creators check a dedicated toggle in YouTube Studio labeled something like "This content contains altered or synthetic material." YouTube then surfaces a disclosure on the video's watch page and in the video player.
Exemptions: Content that is clearly stylized, animated, or uses obvious AI filters (e.g., artistic filters, cartoon-style AI avatars) is generally exempt. Scripts, outlines, or minor AI-assisted edits (like auto-captions) also don't need disclosure.
Penalties: Repeated failure to disclose where required can result in content removal, channel-level warnings, or reduced monetization eligibility.
Meta (Instagram & Facebook)
Effective: April 2024 (announced) — rolling out from May 2024 onward.
What triggers a label: Meta labels content where AI detection is confident — primarily AI-generated images and video uploaded to Instagram, Facebook, and Threads. The label reads "AI info" and appears directly on the post.
How labeling works:
- Automated detection: Meta uses detection models and C2PA metadata to identify AI-generated images and apply the label automatically.
- Creator self-labeling: Creators can also add a "Made with AI" tag manually via the native post options.
- Replaced the old "manipulated media" policy: Meta consolidated its approach in April 2024, moving away from removing content solely on manipulation grounds and instead toward transparent labeling.
Penalties: Content without proper disclosure where it should exist may be deprioritized or removed. For political ads specifically, AI-generated content has carried mandatory disclosure requirements since late 2023.
TikTok
Effective: Policy announced April–May 2024, with full labeling rollout mid-2024.
What triggers a label: TikTok requires creators to label any AI-generated content that contains realistic images, audio, or video — including content that was significantly edited with AI to look real. This is explicitly spelled out in TikTok's Community Guidelines.
How to label: Creators use a native "AI-generated content" toggle within TikTok's posting flow. TikTok also automatically applies labels to content created using TikTok's own in-app AI effects. The platform has partnered with the Coalition for Content Provenance and Authenticity (C2PA) and uses Content Credentials metadata to help identify and label AI content.
What the label looks like: A visible "AI-generated content" label appears on the video. TikTok has been actively enforcing this — unlabelled AI content that should carry the disclosure is subject to removal or reduced reach.
Quick Comparison Table
| Platform | Label Name | Trigger | Method | Exemptions |
|---|---|---|---|---|
| YouTube | Synthetic/AI content toggle | Realistic-looking AI video, audio, or imagery | Creator self-disclosure in Studio | Stylized/animated content, minor AI assistance |
| Instagram / Facebook | "AI info" | AI-detected images/video | Automated + creator self-labeling | Content not altered by AI (per old "manipulated media" rules) |
| TikTok | "AI-generated content" | Realistic AI images, audio, or video | Creator toggle + automatic (TikTok AI effects) | Non-realistic or obvious AI effects |
One Rule That Applies to All Platforms
If you use AI to create content that could deceive your audience — fake news, fabricated testimonials, realistic deepfakes, or synthetic versions of real people — every major platform now requires disclosure. The industry trend is toward automatic detection backed by C2PA/Content Credentials metadata, meaning platforms are getting better at catching AI content even when creators don't self-label.
The bottom line for creators: Check the box or toggle every time you post something a viewer might reasonably believe is real footage but wasn't. When in doubt, disclose — it's easier than dealing with a takedown or content removal.
Try Calabi free at calabilabs.com — 3 cleans, no card.