What Is a Biometric Hash?

Biometric hashing applies a mathematical transformation to biometric data (such as a face embedding) to produce a fixed-length representation that cannot be reversed to reconstruct the original. In the context of NCII scanning, biometric hashing allows platforms to maintain databases of 'known NCII' (such as NCMEC's Take It Down hash database) that can identify matching content without storing the actual intimate images. ScanErase uses biometric embeddings (a related but distinct concept) for its matching process and deletes all biometric data at session end.

Key facts about this term

1. Biometric hashing enables private database matching Hash databases allow platforms to check uploaded content against known NCII without storing or processing the original intimate images — protecting both victim privacy and platform liability.
2. NCMEC's Take It Down program uses this approach for minors NCMEC's hash-sharing program for minors' intimate imagery creates hash representations that participating platforms use to automatically detect and block re-uploads.
3. Biometric hashing differs from file hashing File hashing (like MD5 or SHA-256) creates a hash from the raw file bytes. Biometric hashing creates a hash from biometric features — making it robust to re-encoding, cropping, and minor image modifications.