FAQ · AI content & watermarks

Does cropping remove a watermark?

Short answer

No — cropping only removes a visible corner mark; the C2PA, metadata and encoder signals that get AI content flagged live in the file, not the pixels.

Usually not. Cropping only removes a visible corner watermark, and even then platforms train their detectors on the original signature, so a heavy crop can still fail. More importantly, the signals that actually get AI content flagged — C2PA Content Credentials, XMP DigitalSourceType flags, generator tags and encoder fingerprints — live in the file’s metadata and bitstream, not the pixels, so cropping leaves all of them intact. Invisible pixel-domain watermarks also survive most crops. To post AI content cleanly you need to strip the metadata container and encoder signature, not just trim the frame.

Related on Calabi: Does Instagram detect AI images? · How do I remove C2PA metadata? · Does taking a screenshot remove AI metadata? · How to post AI content on Instagram without getting shadow-banned · Sora — or just open the tool.

Strip it in one pass. 10 free cleans with a forensic proof card.
Try free →
Sources & references