Facebook once built an internal facial-recognition app that allowed employees to identify people by pointing their phone cameras at them.
The app, which was developed by Facebook employees between 2015 and 2016 and tested internally, relied on information from the social network's vast collection of user-uploaded photos and facial-recognition data to identify people in real life within seconds, sources told Business Insider.
It was shown off at an internal all-hands meeting in 2015, one source said, and was presented as an "example of future innovations" that Facebooks' technology and data could enable.
A second source said that in 2016 a beta version of the app was made available for download to Facebook employees that used a smartphone camera to identify anyone for whom the social network had enough data — regardless of whether or not they were a Facebook employee. It was subsequently restricted to identify only people recognized as Facebook friends of the app user (still regardless of whether or not they were an employee), and they used the app on their friends only after that change was made, they said.
A Facebook spokesperson denied that the app could ever identify strangers who didn't work at Facebook, saying that it was able to recognize only the faces of Facebook employees and of Facebook friends of the user who had enabled the social network's facial-recognition system on their accounts.
The app was clearly in its early stages and had a basic camera interface, a source said. When pointed at someone, it would take about three to five seconds to recognize them, and then displayed their Facebook name and profile picture.
For years, Facebook has used facial-recognition technology to identify the subjects of photos uploaded to its core social network and recommend "tagging" them — but the version available to consumers has been applied only to photos and never for real-time real-world face recognition. In September, Facebook said it would change to make facial recognition an opt-in feature, rather than assuming by default that users consent to having their faces scanned.
The existence of the experimental app, which has not been previously reported, highlights Facebook's willingness to experiment with technologies with major societal implications. And it raises questions about the $564 billion company's historic approach to privacy and consent, and its handling of users' data.
A Facebook spokesperson said the company routinely builds experimental apps internally, and that the facial-recognition app has since been discontinued.
"As a way to learn about new technologies, our teams regularly build apps to use internally. The apps described here were only available to Facebook employees, and could only recognize employees and their friends who had face recognition enabled," said the Facebook spokesperson in a prepared statement.
Facebook has a history of privacy scandals
The experimental app was developed at a time when Facebook was under pressure from then rapidly growing rival Snapchat and its comparatively advanced face-tracking technology and filters. A former Facebook employee who worked at the company at the time and was not aware of the beta app speculated that it might have been an experiment by Facebook to better understand the area, in a bid to tackle Snapchat's "looming threat."
Facebook has suffered through more than two years of scandals, from the spread of misinformation and propaganda on its platform in the 2016 US election to issues with its handling of user data. Earlier this year, the company was hit with a $5 billion fine from the Federal Trade Commission after the political-research firm Cambridge Analytica misappropriated more than 80 million users' data as a result of lax controls by Facebook.
Facial recognition has also been a thorny issue for the company. In 2012, Facebook stopped using facial recognition in the European Union over privacy-regulatory issues, before bringing in back in 2018 with stricter controls. The company is facing a lawsuit over allegations that it didn't acquire consent from users in Illinois before using their biometric data, which could cost it up to $35 billion if it loses, Ars Technica reported.
Do you work at Facebook? Contact this reporter via encrypted messaging app Signal at +1 (650) 636-6268 using a nonwork phone, email at rprice@jkmperu.com, Telegram or WeChat at robaeprice, or Twitter DM at @robaeprice. (PR pitches by email only, please.)
Read more:
- Instagram's lax privacy practices let a trusted partner track millions of users' physical locations, secretly save their stories, and flout its rules
- Mark Zuckerberg's personal security chief accused of sexual harassment and making racist remarks about Priscilla Chan by 2 former staffers
- Facebook says it 'unintentionally uploaded' 1.5 million people's email contacts without their consent
- Years of Mark Zuckerberg's old Facebook posts have vanished. The company says it 'mistakenly deleted' them.
