Meta removed the facial recognition code from its smart glasses app a day after WIRED discovered it.

      TL;DR: Meta removed the NameTag facial recognition code from its AI app just a day after WIRED revealed its existence on 50 million phones. Meta claims that no decisions have been finalized.

      On Friday, Meta erased nearly all references to an unreleased facial recognition system from its smart glasses companion app, following a WIRED report that the software had been discreetly integrated into an application used on over 50 million devices. The feature, internally referred to as NameTag, was meant to transform faces captured by the company’s Ray-Ban smart glasses into unique biometric signatures, which would then be compared to a database stored on the user’s device. WIRED also discovered that faces not recognized by the system were cropped, indexed, and stored locally for future processing.

      Andy Stone, Meta’s vice president of communications, told WIRED on Monday that the feature is “purely exploratory” and emphasized that no definitive decision has been made regarding its future. This statement appears inconsistent with the evidence documented by WIRED. The version of Meta AI released the same day as WIRED’s Thursday report contained various code libraries explicitly related to facial recognition, a mechanism for executing the NameTag recognition pipeline, and an alert that would have notified users if an individual was identified.

      The updated version released on Friday removed all of this, as well as a folder designed for storing cropped images and biometric signatures of unrecognized faces. Meta did not respond to WIRED’s inquiries regarding the reasons for the code's removal or whether the modifications were planned prior to the publication of the story. A few remnants remain in the latest update, including a label for an internal debug menu and an inactive link intended to access a recognized person's profile, indicating sections of the system that have been eliminated.

      The divergence between Meta’s public statements and the code found by WIRED is a significant point of tension. Prior to the Thursday article, Stone downplayed the findings, claiming the company couldn't address questions about how the system would function because “the feature does not exist.” Andrew Bosworth, Meta’s chief technology officer, labeled the reporting as “incredibly misleading” and “absolutely dishonest.” However, the code was sufficiently functional to incorporate three AI models—one for face detection, another for cropping, and a third for encoding biometric data—all embedded in the companion app for a product already under scrutiny for privacy concerns.

      Meta did not answer ten questions posed by WIRED before the article's publication, including whether a database of face profiles for NameTag had been created, how long the app retains images and biometric data of unrecognized users, and whether such data would ever be sent to Meta's servers. The company also did not address whether NameTag was being developed for users with visual impairments or respond to critiques from privacy advocates warning that the system could enable stalkers and abusers to identify strangers in public.

      NameTag first emerged in February when The New York Times, citing internal Meta documents, reported that the company was working on facial recognition for its smart glasses and considering a release as soon as this year. An internal memo allegedly mentioned launching the feature during a “dynamic political environment” when privacy and civil liberties advocates would be preoccupied with other issues. WIRED later found that much of NameTag’s framework had been integrated into the Meta AI app as early as January, months before any public acknowledgment, highlighting a pattern of Meta prioritizing product release over transparency regarding its smart glasses.

      Kade Crockford, director of the technology for liberty program at the American Civil Liberties Union of Massachusetts, stated that the removal of the code does not negate the initial decision to implement it, using it as evidence of the need for stronger legal protections for consumer privacy than Congress has provided. Last week, the Massachusetts House of Representatives unanimously passed a consumer privacy bill that, if enacted as proposed, would enforce strong provisions including a private right of action allowing users to sue. “State lawmakers need to do their job and step up to protect consumer privacy,” Crockford asserted.

      Crockford further emphasized that “Meta’s sneaky tactics in embedding the face-recognition code into its smart glasses illustrate why data privacy legislation needs robust enforcement mechanisms.” She added, “Companies like Meta prioritize their profits, so lawmakers must communicate in the only language that resonates with its executives.” Whether the removal of code spurred by investigative journalism signifies a win or just a strategic retreat depends on Meta's future actions and whether regulatory pressures in both Europe and the U.S. lead to significant consequences before the feature quietly reemerges under a different name.