Is the mark invisible, or metadata?
Google Veo marks every video with SynthID, an invisible watermark. Amazon Nova Reel points to provenance metadata that Content Credentials Verify can read unless removed. SceneMixer names a visible badge and a metadata line. From 3 source pages, read 2026-09-25; no maker pays to be listed.
| Source | Figure | Where it applies |
|---|---|---|
| Amazon Nova Reel | Content provenance can be checked with Content Credentials Verify unless the metadata was removed | Nova Reel 1.1, content provenance |
| Google Veo | Every video carries a SynthID watermark that can be verified | Limitations, watermarking |
| SceneMixer | A corner badge and AI-generated-content metadata; always added without a paid plan, off by default on paid plans | The label |
Inclusion rule. Vendor pages and references whose printed figures bear on this delivery target. Order. Alphabetical by source name.
1As printed
Videos created by Veo are watermarked using SynthID
Google AI for Developers, Veo 3.1 in the Gemini API
This should indicate the image was generated unless the metadata has been removed.
Amazon Nova user guide, video generation
2What it means for delivery
An invisible watermark is in the picture and survives re-encoding; metadata sits beside the picture and can be stripped. The two are not the same, and neither is a visible badge.
No figure on watermark on output from 14 vendors: Adobe Firefly, BytePlus Seedance, Flixly, Grok Imagine, Higgsfield, Kling AI, Luma, M Studio, MiniMax, OpenAI Sora, Pexo, PixVerse, Revid AI and Vidu.
- LimitationsEvery video carries a SynthID watermark that can be verifiedwatermarking
- Nova Reel 1.1Content provenance can be checked with Content Credentials Verify unless the metadata was removedcontent provenance
- The labelA corner badge and AI-generated-content metadata; always added without a paid plan, off by default on paid plans