Meta removes NameTag face-recognition code from smart-glasses app

Per reporting by WIRED, code for an unreleased facial-recognition system internally dubbed NameTag was discovered in the companion Meta AI app used with Ray-Ban and Oakley smart glasses. WIRED reported that the code, present in versions of the app installed on more than 50 million phones, would have converted camera images of faces into on-device biometric signatures (faceprints) and that faces the system failed to match were cropped and indexed locally for later processing. Following WIRED's June 4 report, WIRED and other outlets documented that a June 5 update to the app removed nearly all of the NameTag-related code and a folder that would have stored cropped face images and biometric data (WIRED; Engadget; The Next Web).

Per WIRED's code review, the embedded NameTag machinery included distinct processing steps: face detection, face cropping, and encoding faces into numeric biometric signatures; the reporting said multiple AI models were present to support those steps. WIRED and corroborating outlets noted traces left in the updated app - an internal debug label and a dormant link to a recognised-person profile - indicating prior integration points for the feature (WIRED; The Next Web; Engadget).

WIRED reported multiple unanswered questions to Meta ahead of publishing; the outlet and follow-ups state Meta did not respond to a set of queries about why the code had been included or why it was removed. Andy Stone, Meta's vice president of communications, told WIRED the effort was "purely exploratory" and that no "final decision" had been made about the feature; Andrew Bosworth, Meta's chief technology officer, described WIRED's reporting as "incredibly misleading" and "absolutely dishonest," per WIRED and subsequent coverage (WIRED; Engadget; The Next Web).

Editorial analysis: Companies building sensing hardware and companion mobile software have repeatedly faced privacy scrutiny when development artefacts reach production-distributed builds. Observers have previously documented similar instances where dormant or pilot-stage code was present in consumer apps, raising questions about internal release controls, on-device data handling, and auditability. For practitioners, this episode highlights the operational risk of shipping code paths that touch biometric processing even if features are not user-facing.

Editorial analysis: Biometric processing of faces triggers heightened legal and regulatory scrutiny in multiple jurisdictions. Even on-device faceprint creation can attract privacy advocacy and regulator attention because of potential misuse pathways and unclear consent boundaries. Security and privacy engineers implementing camera-based features should expect greater demand for explicit documentation, local-data lifecycle controls, and phased exposure to minimise accidental distribution.

Editorial analysis: Observers should track follow-up reporting and any regulatory inquiries, statements from privacy or civil-rights groups such as the Electronic Frontier Foundation, and whether app-store release notes or developer changelogs document stricter controls. For practitioners, monitoring library provenance, build pipelines, and telemetry that exposes dormant features will be a practical indicator of operational maturity in similar products.