Metadata: easy to read, easy to strip
EXIF, XMP, and C2PA fields sit in the file header. Any image viewer or hex editor can read them, but a single re-save in most apps wipes them out.
Pixel watermarks: harder to add, harder to remove
Pixel-domain watermarks survive a screenshot, a re-encode, and moderate editing. But adding them requires a trained model and detecting them requires a paired key — they're not human-readable.
They're complementary
Robust provenance combines both: cryptographic metadata for verifiable claims when the file is intact, plus a pixel watermark as a fallback for screenshots and re-uploads.
Check AI Watermark
Upload an image to scan for AI watermarks, provenance signatures, and metadata clues.
Open Check AI WatermarkFrequently asked questions
More from this guide
What Is an AI Watermark?
Invisible signals embedded in AI-generated images, and what they actually tell you.
Read articleHow AI Watermarking Works
From pixel-domain perturbations to C2PA manifests — how invisible signals get embedded and read.
Read articleWhy AI Watermarks Exist
Provenance, accountability, and the policy pressure pushing every AI vendor to mark its output.
Read articleDetecting AI-Generated Images
Watermarks, metadata, visual artifacts, and why no single signal is conclusive on its own.
Read article