Facebook Intends to Correct Its Facial Recognition Feature with a De-Identification System

Facebook’s research group, the Facebook AI Research (FAIR) developed a de-identification (De-ID) system as a means to deflect adversarial contentions and intentions against FB’s facial recognition feature.

FAIR sees the De-ID system as a method that would protect the privacy of users, in the event that their photos and videos appear at the social media site, with or without their knowledge and consent. The De-ID system will slightly alter certain facial characteristics to make the facial recognition feature ineffective, and therefore prevent auto-recognition and auto-identification.

 

FAIR is scheduled to introduce the De-ID system that it developed for the Facebook platform, at the week-long International Conference on Computer Vision (ICCV). It will be held in Seoul, South Korea from October 27 2019 (Sun) to November 02, 2019 (Sat.)

FAIR De-ID System Comes a Bit Too Late

The De-ID development comes a bit too late, as it came closely at the heels of Facebook’s failure to stop a class-action lawsuit from proceeding. FB lawyers tried to obtain the support of a higher court via an appeal to stop the multibillion-dollar class-action lawsuit filed against Facebook’s facial recognition practices.

The De-ID system if it had been originally introduced as among the options available to FB users, could have prevented the class action lawsuit filed under the strength of the Illinois Biometric Information Privacy Act. Although a state law, it requires companies operating in Illinois, to institute a public policy prior to collecting and storing biometric data of customers and/or users, including those collected and stored by face-scans. The law also requires laying out of details on how the company stores every biometric data collected.

 

As far as Illinois Facebook users are concerned, they have a right to sue Facebook for privacy violation because of the social media site’s automated facial recognition technology. The system enables others to easily identify a user, by his or her appearance in a photo uploaded and posted by that person or even by another FB user.

Such capability was incorporated as a function of the Tag Suggestions feature, which through the facial recognition system, divulges the identity of every person appearing in publicly posted photos on Facebook.

Facebook attempted to secure a higher court’s support in stopping the proceedings, by arguing that the plaintiffs to the class-action lawsuit have not presented any evidence of concrete injury. Moreover, Facebook’s appeal also contested the lower court’s authority in certifying that cases filed by numerous plaintiffs, qualify as a class-action lawsuit.

However, in a 3-0 decision, the appeals court turned down Facebook’s request and instead gave the lower court authority to continue with the lawsuit proceedings.