WattShed Liveness
Check a photo's capture story before you rely on it.
This reads signals out of a photo's own metadata, not proof. A photo can pass every signal and still not show what someone claims, and it can turn up nothing on a signal for an entirely ordinary reason: messaging apps, some camera settings, and most screenshot flows strip capture metadata before the file ever reaches you. That's reported as absent, never as evidence of fraud.
A capture more than 30 daysold is flagged as stale, and a batch's timestamped photos are expected to fall within 12 hours of each other. An optional location check compares against a 150 m radius by default, adjustable below. Photos are checked in memory for the request and never stored. JPEG originals work best; a re-saved or forwarded copy often carries less metadata than the original file.
Need stronger proof than a photo's metadata? The live capture check records a short video to a one-time pattern this server issues, and verifies the pattern's own execution.
Add photos and run the check to see a report here.
Each photo gets its own set of signals, plus one Session group comparing the batch as a whole.
What's checked
Two groups of signals: one per photo, one across the batch.
Per photo
Readable metadata, capture time freshness, GPS proximity to an expected location (if you provide one), and any recorded processing software.
Across the batch
Whether every photo came from the same device, and whether the timestamped photos fall inside one session.
Each signal reports pass, flag, or absent. A flag means worth a second look, not a finding. Use this alongside your own judgment and, where it matters, a direct conversation with whoever took the photo.