Catch Impersonation Earlier with Reference Photo Mismatch Detection

Identity checks in AI Detections have always compared a candidate's exam photo against the ID photo they take at the start of the session. Now you can anchor that comparison to a photo you trust, not just the one captured on the day.

What's new

You can now upload a reference photo for any candidate from their Media settings. Once it's on file, AI Detections automatically compares the candidate's start-of-exam ID photo against it—once per session—and raises a Reference Photo Mismatch finding if the two don't look like the same person.

Why it matters

The existing identity check only ever compares the exam photo against itself over time—it has no independent source of truth. A reference photo you've verified in advance (from a résumé, an ID document, a prior interview) closes that gap and makes impersonation much harder to slip past.

How it works

• Upload a reference photo from the candidate's Media section—crop, rotate, and resize it right in the editor before saving.
• The photo is checked immediately: a green confirmation means it's usable, a clear red message explains why if it isn't (no face, more than one face, or the face isn't clear enough).
• No extra setup needed—once identity check is enabled on the step and a reference photo exists, the comparison runs automatically on the next completed session.
• A mismatch is treated as strong (red) evidence, just like the in-session identity check.
• The reference photo appears alongside the ID photo and first capture in the session review header, with a red "Mismatch" chip when a finding exists so reviewers can compare side by side.