Provenance Lens

Do ChatGPT Images Have a Watermark?

Learn which provenance signals ChatGPT images carry, how to check their C2PA Content Credentials locally and why a missing signal proves nothing.

Updated: Published by Provenance Lens

The current answer: two provenance layers

OpenAI states that supported images generated with ChatGPT, Codex and the OpenAI API include C2PA Content Credentials and SynthID watermarks. Provenance Lens can inspect the C2PA layer locally. It does not currently implement SynthID detection.

What you can check here

Run the original image through the C2PA checker. A valid OpenAI-issued manifest can establish that the supported export path attached those credentials. The report exposes the signer, assertions and validation status instead of assigning a probability.

Why a screenshot changes the answer

C2PA is metadata and may disappear during screenshots, conversion, editing or sharing. OpenAI uses SynthID as an additional embedded signal, but Provenance Lens does not check it. Missing C2PA therefore does not rule out an OpenAI origin.

What a report cannot prove

  • A valid signature proves who signed a manifest, not that the manifest's claims are true.
  • Metadata is trivially editable. Treat generator fields as an indication, never as proof.
  • Hidden Unicode characters have many innocent causes, including ordinary copy-paste.

Primary sources