Facial recognition search

How does it relate to reverse face search?

The two phrases describe the same operation from different ends. Reverse face search names the user's action: searching backwards from a face to the pages that show it. Facial recognition search names the technology: face recognition applied as a one-to-many search over an index. Search engines and their critics tend to use the second phrase; users searching for the tools tend to use the first.

What makes it different from identification?

A police identification system compares a probe with an enrolled gallery of known people and returns candidate names. A web facial recognition search compares a probe with faces harvested from public pages and returns the pages. The engine attaches no name; a name, if any, comes from the page. The difference is real but narrower than it sounds, because a page often names the person on it, which is why data-protection authorities treat both as biometric identification. See the cases against Clearview AI, whose product is the police version, and against PimEyes, whose product is the consumer version.

Which engines are facial recognition search engines?

Where does it go wrong?

• Look-alikes: the larger the index, the more unrelated faces score high, the doppelganger effect.
• Consent: the people in the index never agreed to be searchable, the issue at the heart of every regulator decision listed on this wiki.
• Misuse: the same search that finds your own photos can be pointed at a stranger, which is why engines publish acceptable-use rules and why several laws restrict the purpose of a search.

Where does the term come up in law?

The EU AI Act prohibits AI systems that create or expand facial recognition databases through untargeted scraping of facial images from the internet or CCTV. The GDPR treats the underlying processing as special-category biometric data. In the United States, Illinois BIPA governs the collection of face geometry regardless of how it is later searched.