We recently learned that Ray-Ban Meta continue recording when you take them off and that there are contractors in Kenya watching everything that happens. They are also being used for things as shady as record women without their consent. In an already delicate context with the issue of privacy, Meta has decided that it was a good idea to add an even more controversial function: facial recognition.
The discovery. Wired has analyzed the Meta AI code, which is the companion-app that comes with the glasses, and they have discovered the code for a facial recognition system that has not yet been officially released. The company had previously said that they were “thinking” about how to implement this technology, but the discovery shows that the code began to be deployed last January. Meta has been quietly including all the pieces necessary to get it up and running, so they could launch it at any time.
How it works. The function is called internally ‘NameTag’, although its commercial name is believed to be ‘Connections’. It combines three AI models: one detects the face through the camera, another scans it, and the third converts it into biometric data. They are then compared to a database on the user’s phone and, if it detects a person, it sends a notification. That is, if someone who is in that database approaches you, they will tell you who they are.
Qwhat does this mean. The ideal scenario from a privacy point of view would be that you can only use it with those contacts who have agreed, but the code says they can take it further. The interesting thing is what happens to faces that are not recognized. The app does not discard them, but rather stores them in a folder called “pending.” And there is something even more striking: the database you have on your mobile is designed to be able to receive Meta updates, that is, they could modify it remotely.
There is a possibility that it was used for an “assistive mode”, for example for blind users, in which Meta could include packets of faces of people who have agreed to be included. However, it also opens the door to the creation of a database of faces, with the risks that entails.
Criticisms and risks. Cooper Quintin, a security researcher consulted by Wired, criticizes that Meta has “created the ability to turn its clients into a distributed surveillance machine.” Additionally, last April, more than 70 organizations and advocacy groups They demanded that Meta abandon the projectarguing that a system like this, integrated into such a discreet wearable, could normalize the silent identification of strangers and poses a risk to especially vulnerable groups such as immigrants or LGBTQ+ people.
Meta has a past. It is not the first time that Meta has had problems with facial recognition functions. In 2010 they integrated photo tagging into Facebook, collecting data from more than 1 billion users. In 2019 the company had to pay a fine of more than 5 billion dollars for privacy violation and in 2021 they closed the system and deleted all facial data. However, Joseph Jerome, a former Meta Reality Labs employee, tells Wired that internally it wasn’t considered a final decision and that “There was always this tension of, well, when do we roll out facial recognition again?”
Meta’s response. A spokesperson for the company has insisted that no facial recognition feature has been launched in its glasses and that it being in the code simply indicates that they are “exploring it”. Furthermore, he adds that if they finally launch it, they will do so “with complete transparency” and insists that “we are not building a central facial database.” We will have to wait for the launch.
Image | Xataka with Gemini
In Xataka | Anduril and Meta have a new and disturbing obsession: turning smart glasses into instruments of war
GIPHY App Key not set. Please check settings