Law-enforcement-sourced identity vs scraped-source search

What the buyer is actually comparing

Scraped-source systems collect photographs from the public internet and use them as a gallery. Law-enforcement-sourced systems start from records that already exist inside the justice system: arrests, warrants, missing-person cases, and other legally defined files.

Those are different products. They have different legal predicates, different error modes, and different answers when a community, a prosecutor, or a defense lawyer asks where a record came from.

Sourcing

UMbRA, the identity database behind Safience, is built from law-enforcement-sourced records. It is not a gallery scraped from social media, yearbooks, or the open web. If a person has not been arrested, convicted, placed on a qualifying registry, named on an active warrant, or enrolled as a missing person through a lawful process, they are not in that corpus.

Scraped-source search can surface people who have never had a law-enforcement event. That is the feature some buyers want. It is also the feature that creates the civil-liberties and discovery problems most agencies cannot defend.

Permissible purpose

Safience deployments run to a lawful purpose: finding a known threat, locating a missing or trafficked person, or enforcing a facility-defined exclusion that already has a legal basis. Facilities receive a human-verified alert. They do not receive a browseable biometric gallery.

A scraped gallery is often used as general identification of whoever walked past a camera. That is population search. Safience is not built as population search.

What happens to everyone else

On Safience sensors, a still image is captured for comparison. Non-matches are deleted. There is no video archive, no plate or vehicle imaging, and no live feed of the public. A trained human reviews every alert before anyone acts.

Where a biometric comparison is required, it happens outside Safience with an independent, NIST-tested provider. Safience does not access, store, or transmit faceprints.

Questions that belong in a procurement

• Where did the gallery come from, and can you produce that provenance in court?
• What is the predicate for putting a person in the gallery?
• What is deleted, and how fast, when there is no match?
• Who reviews an alert before a door is stopped or a person is approached?
• Does the vendor keep video, audio, or a biometric template on the facility?

Legal entity: Biometrica Systems, Inc., doing business as Safience. See About, UMbRA, and Law Enforcement.