ChatGPT and OpenAI image exports can contain file-level provenance such as C2PA Content Credentials plus ordinary creator, software, date, and editing fields. Inspect the exact file, choose what to remove, and download a locally verified copy without uploading it.
The browser inspects and removes supported C2PA/JUMBF containers, EXIF, XMP, IPTC, comments, software fields, and other file-level metadata while preserving the original whenever the format allows.
This page only handles file-level data the browser can safely inspect and remove, so ordinary metadata cleanup never triggers an upload or charge.
Detect supported C2PA containers alongside EXIF, XMP, IPTC, creator, date, software, and comment fields instead of treating every OpenAI export as identical.
Original-mode JPG, PNG, and WebP cleanup removes supported metadata without recompressing the encoded image content.
Review what was found, select removable fields, and see the cleaned file rescanned before a download is offered.
This page removes supported file metadata. OpenAI may also use pixel-embedded provenance signals in some outputs; removing C2PA or metadata does not prove those signals are absent or that the image was not AI-generated.
File formats are messy. These answers describe the supported behavior of this version—not a promise that every trace everywhere has disappeared.
Supported fields include C2PA/JUMBF containers, EXIF, XMP, IPTC, software names, dates, comments, creator information, and other removable file metadata. The exact fields depend on the model, product, format, and export path.
No. This page is a file-metadata cleaner. A pixel-embedded SynthID signal is separate from C2PA and is not removed by deleting metadata containers.
No. Inspection, cleanup, and output verification happen in the current browser. This page does not call OpenAI's provenance API or upload the image.
