NTT DATA · Watermarking Studio
Check any file for AI watermarks — free, private, instant.
Watermarking Studio reads the dual-layer watermark behind AI-generated image, audio, video and text — and gives you a straight answer, not a raw score.
How it works
Upload
Drop in an image, audio clip, video, or text file. Nothing is sent anywhere until you choose to run a check.
Analyze
Watermarking Studio reads the file's signal-level watermark and any signed provenance metadata in parallel.
Verdict
Get a plain-language verdict — watermarked, not watermarked, or inconclusive — with the evidence behind it.
Every kind of content
The same dual-layer watermark works across all four content types — pick one to check a file or mark your own output.
What is dual-layer watermarking?
AI-generated content increasingly carries two independent signals: a robust watermark embedded directly in the pixels, waveform, or token stream, and a separate, signed provenance record describing how the file was produced.
Watermarking Studio's dual-layer approach checks both. The signal watermark is built to survive common edits — resizing, recompression, trimming. The provenance record proves it hasn't been tampered with, for as long as it stays attached.
Detection reads whichever layers are present and returns a plain verdict — no raw scores, no vendor jargon — along with a short explanation of what was found.
For integrators: the marking studio
Embed the same dual-layer watermark into your own output before it ships, so downstream detectors — including this one — can read it later.
Open the marking studio