Meta Removes Facial Recognition Feature from Its Smart Glasses Application Following WIRED Report

A day after WIRED disclosed that Meta had stealthily integrated an unreleased face-recognition system into an app found on over 50 million devices, the company has since taken it down, as reported by a WIRED analysis of the latest version’s code.

The latest iteration of Meta AI, a companion app for its smart glasses, eliminates the inactive software components associated with the system Meta referred to internally as NameTag. The version released on the day of WIRED’s findings included multiple code libraries named specifically for face recognition, but Friday’s update contains none.

Andy Stone, Meta’s VP of communications, informed WIRED on Monday that the feature remains purely experimental, stating: “No final decision has been made on what to do here, if anything.”

On Thursday, WIRED reported that Meta had subtly embedded significant elements of the NameTag system into the Meta AI app. Although never made public, the feature was intended to transform images captured by the glasses into distinct biometric signatures, known as faceprints, and to compare them with a database of faceprints saved on the user’s device. WIRED also discovered that faces that the system couldn’t identify were cropped, indexed, and stored locally for potential future processing.

NameTag was first mentioned in February when The New York Times, citing internal Meta documents, indicated that the company was working on face recognition technology for its smart glasses and considering a launch as soon as this year. One memo allegedly referenced introducing it during a “dynamic political environment,” when privacy and civil liberties advocates might be preoccupied. Last week, WIRED noted that much of NameTag’s infrastructure had already been embedded into the Meta AI app, which millions had downloaded, as early as January, despite Meta’s public stance that it had made no final decision regarding face recognition.

Following WIRED’s report, Stone dismissed the claims, stating the company was unable to provide details on the system’s functionality as “the feature does not exist.” Andrew Bosworth, Meta’s chief technology officer, described the reporting as “incredibly misleading” and “absolutely dishonest.”

Meta did not respond to ten questions posed by WIRED prior to its Thursday publication, which included inquiries about whether the database of face profiles used by NameTag had been established, how long the app keeps photographs and biometric data of unrecognized individuals stored on a user’s device, and whether that data would ever be transferred back to Meta’s servers.

Moreover, Meta did not answer a question regarding whether NameTag was specifically being built for blind or low-vision users and did not address criticisms from privacy advocates who warned that the system could empower stalkers and abusers to identify strangers publicly. Additionally, it did not respond to whether users would have the option to opt in or out of the system.

The newly updated version of Meta AI has removed nearly all signs of the feature Meta claimed was nonexistent. The face-recognition software, along with the code controlling the NameTag recognition process and the “Person recognized” alert that the app would have presented upon identification, has been eliminated. The update also removes a folder designed to store cropped images and biometric signatures of faces captured but not identified.

Meta did not clarify WIRED’s inquiries regarding the reasons for the code’s removal or whether the changes were planned before WIRED’s story went live.