LabelCheck · Local Analysis · V2.0
Upload an image or video to check explicit labels, implicit metadata, and C2PA provenance credentials item by item, with verdicts and fix guidance mapped to regulatory requirements. Files are parsed locally in your browser and never touch a server.
The Measures for Labeling AI-Generated Synthetic Content, effective September 1, 2025, establish a dual-layer obligation of explicit + implicit labels: both layers are mandatory. A closing line like “written with AI assistance” only covers the explicit layer.
Responsibility sits at the publishing and distribution step — even if your generator adds its own watermark, once the label is missing or stripped at publish time, the compliance gap is yours.
Labeled at generation ≠ compliant at publish. Every step between generation and publishing can lose the label — which is exactly why checking exists.
A screenshot is a brand-new file: none of the original metadata (including the C2PA credential) carries over. This is the most common way labels are lost.
Some AI tools’ in-app previews are processed copies; social platforms systematically strip metadata on upload. Always use the “download original” button.
Editor exports, format conversion, and re-compression rewrite the file structure — implicit metadata written earlier is often stripped wholesale at this step.
In practice, AI content with its label removed still publishes normally on most platforms — the system won’t stop you, but the responsibility remains yours.