Remove ChatGPT and DALL·E content credentials
OpenAI attaches a signed C2PA manifest to images generated in ChatGPT and DALL·E. It is a cryptographic record declaring the image was AI-generated, and it is the single marker platforms read most reliably.
This is why a ChatGPT image gets labelled on Instagram almost every time, even when nothing else in the file looks unusual.
writes to: C2PA
Drop an image here
We read what is inside it first, then show you every field before anything is removed.
JPEG · PNG · WebP: up to 25.0 MB · 30 files at a time
What is in the file
What ChatGPT & DALL·E writes into your image
| Field | Contents |
|---|---|
| C2PA manifest | Signed claim naming OpenAI as the generator |
| c2pa.actions | An action list, typically c2pa.created |
| Digital source type | trainedAlgorithmicMedia |
| Claim generator | The OpenAI software and version that produced it |
What it reveals
What someone with your file can learn
- An unambiguous, machine-readable declaration that the image is AI-generated
- The generating software and its version
- Timestamps recording when the image was created
Worth knowing
ChatGPT & DALL·E workflow notes
- Screenshotting a ChatGPT image does remove the C2PA, but it also costs you resolution and adds your screen colour profile. Cleaning the original is strictly better.
- Downloading from the ChatGPT mobile app versus the web app can yield different embedding; check the file you actually have.
- If you edit a ChatGPT image in Photoshop, you often end up with both the OpenAI manifest and an Adobe one.
Questions
ChatGPT & DALL·E metadata FAQ
›Why does my ChatGPT image get labelled on Instagram?
Because OpenAI embeds a signed C2PA manifest, and Meta reads C2PA at upload. It is the most reliably detected marker of all, which is why ChatGPT and DALL·E images are labelled far more consistently than images from tools that only write XMP.
›Does taking a screenshot remove the C2PA data?
It does, because a screenshot is a new file built from pixels on screen. But you lose resolution, gain your display colour profile, and may pick up device metadata. Stripping the manifest from the original keeps full quality.
›Is removing OpenAI content credentials allowed?
Removing metadata from a file you own is ordinarily lawful. Disclosure is a separate obligation: if the context requires you to say an image is AI-generated, that duty stands regardless of what the file contains.
Other tools