Which platforms auto-detect AI media: YouTube, Meta, Pinterest, TikTok

All four vendors describe some automatic labeling, but each means something different, and none says it replaces what you disclose. YouTube can auto-apply a label from its own GenAI tools, C2PA metadata or its detection systems. Meta uses industry-standard signals for ads made with third-party AI tools. Pinterest says its labels follow the IPTC Photo Metadata Standard and its detection systems. TikTok labels content made with its own official AI effects. The safe rule for a clip made with any Sume video model is to disclose it yourself where the platform asks.

Side by side

Each row is that vendor's own wording, condensed.

What this means for Sume output

The Sume docs for video generation describe how to request, poll and download a clip. They do not say Sume embeds C2PA, IPTC or XMP metadata, and they do not describe any platform label. So assume the file arrives without provenance metadata of its own, and do not count on metadata-based detection to label it.

TikTok's page is explicit that official TikTok effects are labeled automatically; a file made elsewhere and uploaded is not described that way, so the disclosure falls to you.

Verdict: disclose manually

Detection is a safety net run by the platform, with no stated accuracy on these pages. Where a platform gives you a toggle or a field, use it. Keep a simple log: clip, tools used, which platforms, and whether you disclosed. Penalties named on the pages (for example removal or Partner Program suspension at YouTube for repeated non-disclosure) are tied to failing to disclose, not to using AI.

• YouTube: use the altered-or-synthetic toggle when content is realistic.
• Meta: check About this ad on live ads.
• Pinterest: flag image Pins you know are AI-made.
• TikTok: use the toggle or an on-screen note for realistic synthetic content.