Free ChatGPT & DALL·E Image Metadata Remover
Images from ChatGPT and DALL·E, whether made in the app or through OpenAI's API, carry a signed Content Credentials (C2PA) record saying they came from OpenAI's tools. This tool shows what is stored in your image and removes it without touching the pixels.
What ChatGPT & DALL·E puts in your images
What we found in a real file: In our own test, a PNG generated with OpenAI's tools contained a single block of metadata: a C2PA manifest stored as its own PNG chunk and signed by OpenAI. It had no EXIF, XMP or prompt text. After cleaning, the pixels were byte-identical to the original.
C2PA Content Credentials
OpenAI began adding C2PA metadata to DALL·E 3 images in early 2024 and says images generated with ChatGPT include it too, with an extra reference to their ChatGPT origin. For API images, it says the metadata includes a signature indicating they were generated with DALL·E 3.
Nothing about your prompt
In our test file there was no prompt text or generation settings. The metadata states where the image came from, not what you typed. It is stored in one block, so it can be removed without re-encoding the image.
What this tool can’t remove
OpenAI also says images from ChatGPT and its API carry a SynthID invisible watermark, and its openai.com/verify tool checks for both that watermark and the C2PA record. The watermark lives in the pixels, so this tool cannot remove it. Cleaning the metadata removes the Content Credentials, but it does not make an image undetectable by OpenAI's own verification.
Sources: OpenAI: Provenance signals in OpenAI-generated content · OpenAI: Verify OpenAI-generated content · The Decoder: OpenAI equips DALL·E 3 with C2PA metadata
How to remove metadata from a ChatGPT & DALL·E image
- 1
Save the original file: download it from ChatGPT, or save it from the API response. Don't screenshot it, because a screenshot has no metadata to remove.
- 2
Drop the file into the tool above. It is read in your browser and never uploaded.
- 3
Review the C2PA item it finds. Tick “Keep” only if you want to leave it in.
- 4
Choose “Remove all metadata”, then download the clean copy and the optional report of what was removed.
ChatGPT & DALL·E metadata questions
Do ChatGPT and DALL·E add metadata to images?
Yes. OpenAI says images generated with ChatGPT and DALL·E 3 include C2PA Content Credentials. In our test, a PNG generated with OpenAI's tools held a single signed C2PA manifest.
When did DALL·E start adding C2PA metadata?
OpenAI announced C2PA support for DALL·E 3 images in early 2024. Older images made before that may not contain it.
Will removing the metadata hide that my image came from OpenAI?
Only partly. It removes the Content Credentials, but OpenAI also says its images carry an invisible SynthID watermark in the pixels. OpenAI's verification tool can check for that watermark, and this tool does not remove it.
Do I need to remove metadata if I screenshot the image?
A screenshot normally does not carry the original file's metadata, so there is usually nothing left to remove. OpenAI notes that screenshots and many social platforms strip this data. The pixel watermark is not affected by either.
Does cleaning change the image quality?
No. The pixel data is copied exactly as it was and only the metadata block is dropped, so the picture is identical. The file gets slightly smaller.
Is my image uploaded anywhere?
No. The file is processed on your device in your browser, and the cleaned copy is created locally.
Other AI image generators
- Midjourney image metadata remover
- Stable Diffusion image metadata remover
- Nano Banana image metadata remover
Or go back to the general AI image metadata remover.