What Meta’s Hidden Facial Recognition System Is and Why It Matters
Meta’s hidden facial recognition system is an unreleased feature embedded in the Meta AI companion app that turns faces seen through smart glasses into biometric identifiers stored and matched directly on a user’s phone, raising serious concerns about smart glasses privacy, biometric data security, and everyday surveillance technology. Investigators at WIRED say they found references in the live app to an internal feature called NameTag, designed to identify people captured by Ray-Ban and Oakley smart glasses. The app has been downloaded more than 50 million times, meaning the Meta facial recognition code has already reached a huge user base even though it is not enabled for consumers. Core components reportedly landed in January, including three AI models that detect faces, crop them, and convert them into biometric data that can be checked against faceprints stored on the device, then used to alert the wearer when someone is recognized.
How the Code Works: From Faces to “Faceprints” on Your Phone
EFF’s Threat Lab confirmed through static analysis that Meta’s app contains active facial recognition code, even though the feature is hidden from the normal interface. According to the Electronic Frontier Foundation, the system stores “faceprints as a series of 2,048 numbers uniquely representing the positioning of a person’s facial features.” When enabled, every new face in the glasses’ field of view can be turned into this numerical template and compared against all existing faceprints in the local database. Another researcher found that by connecting a phone in debug mode and manually adding a face to the app’s database, the glasses then detected that face when it appeared, proving the pipeline works end to end. While Meta says it is not building a central face database, on-device matching still creates a powerful, portable surveillance technology tied to your daily life.
Smart Glasses Privacy: Always-On Cameras Meet Biometric IDs
Ray-Ban Meta glasses already raise alarms because modders have shown how to disable the recording LED, enabling covert filming. Adding Meta facial recognition capability turns that always-on camera into a biometric scanner. Even if Meta limits matching to people you have added as “Connections,” the same biometric data security risks remain: anyone you add has their face turned into a durable identifier. A May app version appears to rebrand NameTag as “Connections,” inviting users to “remember the people you met,” which sounds helpful but glosses over the surveillance implications. Once faceprints exist on your phone, they could be exposed through device compromise, misuse by someone who gains access, or future feature changes that expand who or what the system can recognize. Hidden, pre-deployed code also means users never had a clear chance to consent before this surveillance technology arrived on their devices.
Meta’s Track Record and the Risk of a Distributed Surveillance Network
EFF warns that Meta’s design “created the capacity to turn their customers into a distributed surveillance machine,” where millions of wearers quietly scan and catalog the people around them. Meta previously shut down its Facebook face-recognition system following criticism, and it has already paid USD 650 million (approx. RM2,990,000,000) to settle a BIPA lawsuit over mass tagging of photos based on facial recognition. Internal documents reported earlier described Meta’s interest in launching facial recognition for smart glasses “during a dynamic political environment” when critical civil society groups might be distracted, suggesting the company understands how controversial this surveillance technology is. Even without a central database, linking smart glasses privacy issues to persistent biometric identifiers risks normalizing face scanning in everyday spaces, from workplaces to social events, in ways that bystanders cannot easily detect or refuse.
What Users Can Do Now to Protect Their Biometric Privacy
For now, Meta says no facial recognition feature has shipped to consumers and promises full transparency before any release. Yet the code is already present and confirmed active under the hood, so users concerned about biometric data security should act as if the capability could appear with a future update. If you own Ray-Ban or Oakley Meta glasses, consider disabling or uninstalling the Meta AI companion app when not needed, and review app permissions for camera, storage, and network access. Treat any future “Connections” or “remember the people you met” options as equivalent to building a personal face database and weigh whether you want that responsibility. For everyone else, think twice before appearing on camera around smart glasses and before buying such devices yourself; once facial recognition becomes normalized in consumer wearables, rolling back this level of surveillance technology will be far harder.
