Photo checks for insurers and claims processing
Insurers can check claim photos for signs of generation or editing before an adjuster spends time on them. PhotoCheck flags AI-generated damage photos, reads the camera EXIF and capture time, and reports C2PA records of AI edits, all in one call per photo.
The problem
Generative editing makes it trivial to add a dent, a crack or water damage to a real photo. The edited file often keeps the original camera EXIF, which is why the AI score and the C2PA edit history matter more here than camera data alone.
The checks that matter
| AI score | Flags images that read as generated. |
|---|---|
| C2PA composite | Editors that record AI steps mark the image as a composite with AI content. |
| Capture time | A photo taken before the claimed date of loss is worth a question. |
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(claimPhoto);
const taken = r.metadata.exif.taken_at && new Date(r.metadata.exif.taken_at);
if (taken && taken < dateOfLoss) flag("predates_loss");
if (r.summary.ai_generated === "declared") flag("ai_edit_declared");
if (r.ai.score >= 0.8) flag("likely_synthetic");Questions
Can it find a small AI edit in a real photo?
The overall score often can't, because most pixels are real. A C2PA composite flag can, when the editor recorded it.
Are claim photos stored?
No. They're processed in memory and discarded.