acm-header
Sign In

Communications of the ACM

ACM TechNews

Facial Recognition's 'Dirty Little Secret': Millions of Online Photos Scraped Without Consent


View as: Print Mobile App Share:
Training artificial intelligence on stolen images.

Legal experts warn people's online photos are being used without permission to power facial-recognition technology that could eventually be used for surveillance.

Credit: Greg Peverill-Conti

Legal experts warn people's online photos are being used without permission to power facial-recognition technology that could eventually be used for surveillance.

Said New York University School of Law's Jason Schultz, "This is the dirty little secret of [artificial intelligence] training sets. Researchers often just grab whatever images are available in the wild."

IBM recently issued a set of nearly 1 million photos culled from the image-hosting site Flickr, and programmed to describe subjects' appearance, allegedly to help reduce bias in facial recognition; although IBM said Flickr users can opt out of the database, deleting photos is almost impossible.

Some experts claim these projects not only infringe on personal privacy, but also raise concerns that law enforcement will use facial recognition to unfairly target minorities.

Northeastern University's Woody Hartzog said, "Facial recognition can be incredibly harmful when it's inaccurate and incredibly oppressive the more accurate it gets."

From NBC News
View Full Article

 

Abstracts Copyright © 2019 SmithBucklin, Washington, DC, USA


 

No entries found

Sign In for Full Access
» Forgot Password? » Create an ACM Web Account