Startup wants to give biometric face scrapers digital dead ends to chase

Startup wants to give biometric face scrapers digital dead ends to chase

A small shop of AI coders and photographers have a new use for their GAN-made portfolio of faces. The company says its Anonymizer app creates red herrings to throw face scrapers off a person’s online trail.

Generated Media Inc.’s Anonymizer digests a head shot submitted by a person and creates a seemingly infinite number of digital lookalikes (or maybe look-similars) free of charge. People can find one they like and post it as a profile photo on social media, for example.

The company sells synthetic head shots for use in advertising, marketing and other business roles.

If the company’s Anonymizer works as designed, the artificial photos look enough like the person that family and friends see a strong resemblance but are not fooled, as would happen with a caricature.

The generated images are dead ends for face scrapers like those loosed online by Clearview AI to supply references for facial recognition. They do not belong to any actual human. (An argument could be made that a generated image might very closely resemble someone else unrelated to the transaction.)

By contemporary standards, the company’s data sets are small, and that limits the degree of similarity for some people using the Anonymizer.

Texas-based Generated Media uses two data sets. One, its synthetic collection, is 2.7 million facial images. The second is comprised of almost 100,000 images of real people. The company says its algorithms were trained on studio images from stock-photo shop Icons8.

The whole endeavor has the feel of a startup looking for spaghetti to throw. An article about the Anonymizer talks about how it could be used on dating sites to coyly entice responses without outright catfishing someone. That one seems like a nuance that could easily be lost on the unsuspecting.

One flaw in the company’s business model is the evitable day when Clearview AI and its brethren deploy machine learning that sniffs out fakes. That is, after all, one of the tools being discussed for spotting deep fakes.

Related Posts

Article Topics

 |   |   |   |   |   | 

Comments

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Most Read This Week

Featured Company

Biometrics Research

Biometrics White Papers

Biometrics Events

Explaining Biometrics