Meta Embedded Face Recognition in Smart Glasses App Code

Meta has embedded functional face-recognition technology into the companion app required for its Ray-Ban and Oakley smart glasses, according to a technical analysis conducted by WIRED and verified by independent security researchers.

The feature, internally designated "NameTag," includes AI models capable of detecting faces captured by the glasses' camera, converting them into biometric signatures, and matching them against a database of stored faceprints. While not yet activated for users, the core components have been present in the Meta AI app since at least January 2025, even as the company publicly described face recognition as something it was "thinking through."

The Meta AI companion app, which has been downloaded more than 50 million times, is necessary for operating key features of Meta's smart glasses. Code analysis reveals three AI models already deployed to users' phones: one for face detection, one for cropping, and a third for encoding faces into biometric data.

How the system works

When activated, NameTag would process faces captured by smart glasses cameras and generate unique biometric signatures. These would be checked against faceprints stored locally on the user's device—a database configured to receive updates from Meta's servers. Recognized faces would trigger notifications to the wearer, while unrecognized faces would be indexed and saved to a "pending" folder.

Recent versions of the app rebrand the feature as "Connections" and invite users to "remember the people you met." However, Meta has not disclosed whose faces would be included in the recognition database or how those profiles would be created.

Independent security researcher Buchodi successfully tested the recognition pipeline by adding a single faceprint to the app's gallery. The system correctly identified the image and generated a "Person recognized" notification, confirming the matching functionality is operational.

Why it matters

This development represents a significant shift in Meta's approach to facial recognition technology. The company deleted over a billion faceprints in 2021 following years of controversy and paid $650 million to settle an Illinois class-action lawsuit over its photo-tagging system. In 2024, Meta agreed to a separate $1.4 billion settlement with Texas over allegations of unlawful biometric data collection.

The deployment of face recognition in a mass-market wearable device raises distinct privacy concerns. Unlike smartphone apps that require users to actively point cameras at subjects, smart glasses enable passive, continuous capture of faces in public spaces. More than 70 advocacy groups, including the ACLU and Electronic Frontier Foundation, have warned that such technology could enable stalking and abuse by allowing silent identification of strangers.

"Despite the billions of reasons not to, Meta seems to have created the capacity to turn their customers into a distributed surveillance machine," says Cooper Quintin, a security researcher with the Electronic Frontier Foundation who independently verified WIRED's findings.

Meta spokesperson Ryan Daniels stated that the company is "exploring these types of features" and emphasized that "nothing has shipped to consumers." He added that Meta is "not building a central face database" and that any eventual rollout would be done "with full transparency."

WIRED's code analysis shows the system is currently designed to pull faceprints from Meta's servers and store them on individual user devices rather than maintaining a centralized database.

These details were first reported by WIRED.