How to Remove ChatGPT and DALL·E Image Metadata
ChatGPT image exports and DALL·E outputs can embed C2PA Content Credentials and related AI metadata. Strip those signals locally before posting to social apps.
What ChatGPT / DALL·E often leave in the file
OpenAI image workflows commonly write provenance-style metadata so downstream apps can see that a generative tool was involved. That can include C2PA Content Credentials and related container fields depending on export format and client.
When you download an image from ChatGPT and upload it straight to Instagram or LinkedIn, those platforms may read that metadata and show an AI Info, Made with AI, or Content Credentials style disclosure. See Instagram AI Info and LinkedIn CR badge.
Clean the download before you share
Save the image as PNG, JPEG, or WebP if your client allows it, then strip removable metadata locally.
- Download the image from ChatGPT (avoid relying on a compressed social screenshot).
- Drop it into Unmade AI to remove removable C2PA, EXIF, and XMP.
- Upload the cleaned file to the destination app.
Batch tip for carousels
If you are posting multiple ChatGPT slides, select a folder or multi-select files in Unmade AI, process the batch, and download a ZIP of cleaned copies so every slide in the carousel starts from a cleaned file.
Limits to remember
Unmade AI targets removable metadata. It does not claim to defeat every future classifier or pixel-level watermark. Prefer supported formats (PNG, JPEG, WebP) for the cleanest results today.