Clearview AI Is Running Facial Recognition on 3B+ Scraped Photos

Over the weekend, the NYT profiled Clearview AI. The Peter Thiel-backed startup runs a facial recognition algorithm on a database of what it says are 3+ billion scraped photos. You’ve got questions, I’ve got answers.

Where do those photos come from? Facebook, YouTube, Twitter, Instagram, Venmo, and "millions of other websites." Plus, government-issued driver's licenses and mug shots.

• Pro tip: If you use Venmo, set your interactions to private so third parties can't scrape your info from the public feed.

Who created the app? Hoan Ton-That, a 31-year-old entrepreneur whose prior claim to fame was an app that let users put President Trump's hairdo on their own photos.

Clearview’s clientele? Law enforcement. Sources told the NYT that Clearview's tool is compelling because it works on faces shot from different camera angles.

• The National Institute of Standards and Technology hasn't vetted the company's claim that it's able to find matches up to 75% of the time.

What do experts think? "Absent a very strong federal privacy law, we're all screwed," Stanford privacy professor Al Gidari told the NYT.