A signal, not a stamp
An AI watermark is a signal — visible or hidden — that an image was produced or processed by an AI model. Modern watermarks are usually invisible: subtle pixel patterns, frequency-domain modulations, or metadata fields like C2PA content credentials.
Two broad kinds
Pixel-level watermarks are baked into the image itself and survive some editing. Metadata watermarks live in the file's header (EXIF, XMP, C2PA manifests) and are easy to strip but easy to read.
What a watermark proves
A detected watermark is strong evidence the image touched an AI pipeline. Its absence proves nothing — many generators don't watermark at all, and metadata is routinely removed when images are re-saved or screenshotted.
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
How 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 articleAI Watermarks vs Metadata
Embedded pixel signals and EXIF/C2PA manifests solve overlapping problems in different ways.
Read articleDetecting AI-Generated Images
Watermarks, metadata, visual artifacts, and why no single signal is conclusive on its own.
Read article