Free AI Image
Deepfake Detector

Your file is not kept. It exists only for as long as the scan takes, and it is deleted the moment we hand back a result. No backups of the image, no quiet copy kept for training. We could not show you a past upload even if you asked, because one does not exist. What can be kept is the result: a report you choose to share holds the verdict and evidence, not your file.

To check a file, upload a photo (JPG, PNG, WebP or HEIC) or a video (MP4 or MOV, up to 100 MB). Magic Cleanup takes photos, and the cleaned image comes back as a JPG or PNG at the quality you put in. We don't re-compress your export into a blurry mess on the way out.

No detector is right every single time, and we won't pretend ours is. The number you see is a confidence reading, measured against millions of real photos and generated images from models like Midjourney and Flux. Read it as strong evidence to act on, not a courtroom verdict.

Up to three checks run in order. First the file's own record: content credentials and metadata that can name the tool that made it. Then the invisible watermark OpenAI and Google put in their AI images. Then a detection model scores the pixels. The first decisive answer wins, and the report shows which one it was.

Clone tools just smear nearby pixels over the gap. Ours rebuilds it with a diffusion model that reasons about light, edges, and texture, so the patch actually belongs to the scene instead of floating on top of it.

Resolution stays put. You get the original dimensions back, up to 4K, with no downsampling and no blur. There's no quality tax for running the cleanup.