Photo checks for marketplaces and classifieds
Marketplaces can run every listing photo through PhotoCheck to catch AI-generated product images and photos of items that don't exist. The AI score flags rendered or generated product shots, and the metadata check shows whether a photo came from a phone camera or an image generator.
The problem
Scam listings increasingly use generated photos: a flawless used sofa in a sunlit flat, a car with no plates in a too-perfect street. Sellers of real items photograph them on phones, which leave camera EXIF behind. Generated listings leave none, or name their generator.
The checks that matter
| AI score | Flags generated and rendered product shots for a human look. |
|---|---|
| Camera EXIF | A phone make and model is weak evidence a real camera took the picture, if your upload path keeps EXIF. |
| Faces | Find listing photos with people in them, to blur or reject under your own rules. |
A decision rule
Built on POST /api/v1/analyze. Thresholds are starting points; tune them to what a mistake costs you.
const r = await check(listingPhoto);
const camera = r.metadata.exif.make !== null;
if (r.summary.ai_generated === "declared") return "hold";
if (r.ai.score >= 0.8 && !camera) return "hold";
return "publish";Questions
Our app strips EXIF before upload. Is metadata still useful?
Call PhotoCheck before your pipeline strips it, on the original file. After stripping, only the AI score remains.
What about real photos with a background removed?
Heavy editing can raise the score. Use the score as a reason to review, not to remove.