Photo checks for journalists and fact-checkers
Fact-checkers can use PhotoCheck as a first pass on a viral image: does it read as AI-generated, does the file carry C2PA Content Credentials or camera EXIF, and when and on what device was it taken? It won't settle a story by itself, but it can tell you in seconds which leads deserve the next hour.
The problem
A breaking-news image arrives as a screenshot of a screenshot. Most provenance is gone, and generators now produce convincing news photography. The first questions are always the same: is there any metadata left, and does the image itself look generated?
The checks that matter
| Camera EXIF | Make, model, timestamp and GPS, when the original file is available. |
|---|---|
| C2PA | Some newsroom cameras and editing tools now sign their output. |
| AI score | A quick read on whether the pixels look generated. |
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(viralImage);
report({
camera: r.metadata.exif.model,
taken: r.metadata.exif.taken_at,
where: r.metadata.exif.gps,
credentials: r.metadata.c2pa.present,
ai: r.ai.score,
});Questions
Can it prove an image is real?
No tool can. A low AI score and intact camera EXIF are consistent with a real photo, and both can be faked.
Is the GPS location returned?
Yes, when the file contains it. Nothing is stored on our side.