AIWatermarkCheckerOpen tool

Free · browser-side · no uploads

Watermark checker

Drop an image and see exactly which AI watermarks and provenance records it carries — signed content credentials, generator claims, action history, metadata leftovers. Private by design: everything runs in your browser.

Runs entirely in your browser — your file is never uploaded.

Drop an image here, or browse files

JPEG · PNG · WebP — up to 25 MB. Files are read locally and never uploaded.

Try a sample:

Quick answer

There are two honest ways to “check a watermark” in 2026: read the publicly-verifiable metadata layers (C2PA content credentials — readable by anyone, right here), or use the vendor's own gated detector for invisible pixel watermarks like SynthID. This page does the former and links the latter — no fake detections, no accusations, just receipts.

Which mark will this checker find?

MarkWhere it livesReadable here?
C2PA Content CredentialsSigned metadata store inside JPEG/PNG/WebP (JUMBF)✔ fully — this is our core reader
Generator claims & action historyInside the C2PA manifest (CBOR)✔ — listed per file with timestamps
EXIF/XMP vendor stringsPlain metadata text (editors, “AI-generated” tags)✔ as soft hints only
SynthID (invisible pixels)Image pixels themselves✘ — Google's portals only; we link you there

Check a specific vendor's output

Claude watermark checkerClaude file credentials + the truth about Claude's text watermarkChatGPT / GPT image detectorRead C2PA metadata from GPT Image & DALL·E outputsGemini & SynthIDWhere SynthID can officially be verified, and how to read the C2PA layerGot text instead?Scan for zero-width and invisible characters

FAQ

What is an AI watermark checker?＋

It's a tool that looks for provenance marks inside files — not pixel guesses. Modern AI products leave checkable traces: OpenAI embeds C2PA metadata in image generations, Claude attaches signed content credentials to files, Google pairs SynthID pixels with C2PA metadata. This checker reads the layers that are publicly verifiable, in your browser.

How do I check if an image is AI-generated?＋

The most reliable check is provenance: drop the original file above and read its metadata. Signed credentials naming Claude, ChatGPT, Firefly, or Google models are strong evidence. If nothing is found, be careful: screenshots and image compression strip metadata, and some AI tools never embedded marks — so a clean result does not prove a human made it.

Does this watermark checker work on text?＋

File-based watermarks only here. AI text watermarks (Claude's SynthID-Text-derived mark, Gemini's) are cryptographic and vendor-keyed — see our Claude and Gemini pages for what can and cannot be checked, plus our hidden-character scanner for text.

Is checking a watermark legal / free?＋

Reading metadata in your own browser is free and private by design — files never leave your device. It's evidence for your own decisions, not legal advice.

Why does my check come back empty for a clearly-AI image?＋

Common causes: it was screenshotted, re-saved, or processed by a platform that strips metadata; the tool (e.g. Midjourney) never embedded any; or the pixel-level watermark (SynthID) is present but only detectable through Google's official portals — metadata and pixels are independent layers.

Sources