What Was Discovered Inside Meta’s Smart Glasses App?
Meta’s facial recognition smart glasses controversy centers on dormant code in its companion app that can detect faces, create biometric fingerprints, and identify people around the wearer without their knowledge or consent. Researchers who examined the Meta AI app, which supports Ray-Ban and Oakley smart glasses, found an internal feature called NameTag designed to recognize people seen through the glasses’ camera. Three on-device AI models detect, crop, and convert faces into 2,048‑dimensional biometric data, then compare them against a local faceprint database. Although NameTag is not exposed in the interface, the code has appeared in multiple updates since January and can be triggered through developer tools. One test even produced a “Person recognized” notification after pre-loading a face template, showing that the entire recognition chain already works end-to-end. This hidden pipeline moves the product much closer to smart glasses surveillance than most users realize.
How the Hidden NameTag System Works on Your Phone
Under the hood, the Meta AI app (internally called Stella) contains a full facial recognition pipeline, not a simple camera helper. It ships three ExecuTorch models: SCRFD for face detection, KPSAligner for aligning facial landmarks, and a scaled-up SFace variant that turns each face into a 2,048‑number biometric fingerprint. These embeddings are stored in an SQLite vector database on the phone for similarity matching, along with cropped face images in a private directory that survives normal app use. When enabled through debugging, the app converts each new face in view of the glasses into an embedding and compares it to the stored biometric data collection. If it finds a match, a notification such as “Person recognized” can be sent to the wearer. During controlled tests, researchers even got the system to recognize a portrait of Michel Foucault once his biometric template was added, confirming that the pipeline is ready for real-world use.
Meta’s Official Line vs. Its Track Record on Face Data
Meta says the embedded NameTag code is experimental and inactive. A company spokesperson told reporters the findings are “merely evidence” that Meta is exploring these features and that “nothing has shipped to consumers and no final decision has been made.” Meta also insists it is not building a central face database and promises transparency if the feature launches. Yet this sits uneasily beside its history. In 2021, Meta shut down Facebook’s facial recognition system and deleted over a billion faceprints after regulatory pressure and major biometric privacy lawsuits. The company has already paid USD 650 million (approx. RM2,990 million) to settle one BIPA case and later agreed to a USD 1.4 billion (approx. RM6,440 million) settlement in another. For privacy advocates, embedding a working recognition system into an app installed tens of millions of times looks less like harmless experimentation and more like groundwork for a distributed surveillance network.
Why Always-On Smart Glasses Surveillance Worries Privacy Experts
Facial recognition smart glasses combine three risky elements: always-on cameras, invisible data processing, and biometric data collection that can’t be changed if leaked or abused. The Electronic Frontier Foundation warns that this turns wearers into a “distributed surveillance machine,” since every person in the glasses’ field of view could be scanned and converted into faceprints. Hidden or disabled recording LEDs, which modders have already shown is possible, make it even harder for bystanders to know they are being recorded or analyzed. Unlike unlocking your own phone with your face, NameTag is built to recognize other people who may never have consented. If social norms shift and such features become common, walking down a street, entering a store, or attending a protest could mean being silently logged, recognized, and remembered by dozens of devices without any clear way to opt out.
What Users Should Do Now About Meta Privacy Concerns
Even while NameTag stays hidden, users should treat Meta’s smart glasses as always-on surveillance devices. If you own Ray-Ban Meta or similar models, review the app’s permissions, disable cloud backups you do not need, and think carefully before linking contacts or photos that could seed a recognition database. When wearing the glasses, avoid pointing them at people in private or sensitive settings and explain to friends or coworkers when the camera is active. If you are uncomfortable with the idea of your own face being captured, limit how close you stand to someone using these devices and ask them to remove or power down the glasses in intimate spaces. For many, the safest option is to “think twice before buying or using Meta’s surveillance glasses,” as the EFF puts it. Until stronger rules on biometric surveillance arrive, individual choices remain the main defense.
