What Is Biometric Scanning?

Traditional NCII detection methods rely on file hash matching (identifying known copies of the same file) or keyword search (searching for names or descriptions). Both methods fail for NCII: perpetrators change file names and re-encode content to defeat hash matching, and victims' names may never appear in metadata or descriptions. Biometric scanning addresses these gaps by extracting the mathematical representation of a person's facial geometry from a reference photo and searching for that geometric signature across indexed content — finding re-encoded copies, AI-generated deepfakes, and content posted without the victim's name.

Key facts about this term

1. Biometric scanning uses your face geometry, not your name ScanErase extracts a unique facial embedding from your reference photo — a mathematical representation of your facial geometry — and compares it against billions of indexed embeddings.
2. It finds re-encoded and AI-modified content Re-encoding, compression, and AI modification change the file but preserve the facial geometry. Biometric scanning detects these copies that hash-matching systems miss.
3. Your reference photo is processed in memory and deleted ScanErase does not store biometric data. Your reference photo and the facial embedding extracted from it are processed in memory during the scan and deleted at session end.