Is that photo a real capture, or AI?
AI images are now good enough to fool the eye, so the eye is no longer enough. The clues are in the file's hidden metadata. Upload an image and this free tool reads that metadata, the EXIF a camera writes and the XMP that AI tools and editors write, and tells you whether it looks like a genuine photo, an AI-generated image, or something that has been edited. It all happens in your browser, so the image never leaves your device.
What it looks at
- Camera fingerprints: make, model, lens, capture date and GPS. AI images do not have these; real photos usually do.
- AI markers: a generator name (Midjourney, DALL-E, Firefly and others) or the standard "trained algorithmic media" label that AI tools increasingly embed.
- Editing software: Photoshop, Lightroom, GIMP and similar, which show the pixels were processed.
Why "no metadata" is not a verdict
The single most important thing to understand: an absence of metadata does not mean an image is AI. Screenshots have no camera data, and every major social network strips metadata on upload, so a genuine photo saved from Instagram, WhatsApp or X will look bare. This tool treats missing metadata as inconclusive, not as proof of anything, and says so clearly. It is honest about what the evidence can and cannot show.
Metadata is evidence, not proof
Metadata can be edited or removed, so treat the verdict as strong evidence rather than certainty. Camera fields can be copied onto an AI image, and AI labels can be stripped. For a signal that is much harder to fake, cryptographic Content Credentials (C2PA) are embedded by some AI tools and cameras; our Content Credentials Inspector reads those. When it really matters, combine this check with a reverse image search to see where else the picture appears.
Private by design
The image is read entirely in your browser to pull out its metadata, and nothing is uploaded. This is part of the RealCheck set of tools for the AI era, helping you answer one question: is this real?