Remove AI Label — check & strip AI metadata from images

No — and nothing can. Platforms apply that label themselves at posting time from their own detection plus whatever credentials the file carried. What you control is the before: check and clean a file before uploading so you don't leak prompts, workflows or location, and so the file doesn't carry credentials you didn't intend to publish. After posting, only the platform controls the label. Tools that claim otherwise are selling something they can't deliver.

Detected: C2PA/JUMBF content credentials (including the standard “trainedAlgorithmicMedia” source-type marker), XMP fields naming generative tools (Firefly, Midjourney, DALL·E, Stable Diffusion, Flux, ComfyUI…), EXIF Software/UserComment strings, GPS presence, IPTC blocks and file comments. A real camera Make/Model is shown too — it's a signal the photo is original.

Stripped: all of the above, by rebuilding the file without those blocks. ICC color profiles are kept and pixels are never re-encoded, so the image renders identically.

No. Parsing, cleaning and hashing all run in your browser with the standard File and WebCrypto APIs. Nothing leaves your device; closing the tab erases everything.

Privacy (GPS and device details), confidentiality (many AI tools embed your prompt and workflow into the file), and deliberate disclosure — if you must disclose AI use under a platform policy or regulation, do it in your post, not by leaking it from a metadata field. Cleaning does not remove your obligation to follow disclosure rules.