Security & Riskmetaface recognitionprivacysmart glasses

Meta Removes Face-Recognition Code From Smart Glasses

|June 8, 2026

6.8

Relevance Score

Meta Removes Face-Recognition Code From Smart Glasses — Photo: gizmodo.com · rights & takedowns

Reporting by WIRED identified an unreleased face-recognition feature, called "NameTag," embedded in the code of the Meta AI companion app, which WIRED says is installed on more than 50 million devices. WIRED and Gizmodo report the code was not enabled and exploratory, and that Meta removed the face-recognition libraries from the most recent Meta AI app build after the story surfaced. Gizmodo reports Meta communications VP Andy Stone called the coverage "intellectually dishonest" and "pure advocacy-driven click bait." Reporting also notes Meta has explored facial-recognition ideas for its Ray-Ban smart glasses since at least 2021, per Gibmodo. Editorial analysis: Industry observers will see this as another high-profile example of tension between product exploration and privacy expectations for wearable devices.

What happened

Reporting by WIRED uncovered an unreleased face-recognition system hidden in the code of the Meta AI companion app, which WIRED says is installed on more than 50 million phones. WIRED's analysis identified the feature, reported as named "NameTag," as embedded but not enabled in the shipped app. Gizmodo and WIRED report that, after the reporting, Meta removed the face-recognition libraries from the latest version of the Meta AI app available for download. Gizmodo reports a post from Meta communications VP Andy Stone that described the coverage as "intellectually dishonest" and "pure advocacy-driven click bait." WIRED notes Meta has not provided a detailed explanation of why the code was present or whether the feature will return.

Editorial analysis - technical context

Facial-recognition on wearable cameras raises a tightly coupled set of technical tradeoffs that are common across vendors. Systems that identify people in video generally require (a) a face-detection and alignment pipeline, (b) a face-embedding model, and (c) a reference database for identity matching. Deploying the matching stage on-device reduces continuous network exposure but increases local storage and compute requirements. Offloading matching to cloud services reduces device cost but introduces telemetry, latency, and legal complexity. False positives, threshold tuning, and dataset bias are recurring failure modes that magnify privacy and safety concerns in always-on or opportunistic capture contexts.

Industry context

Gizmodo and WIRED place this incident in a longer timeline of Meta exploring facial-recognition for smart glasses, citing earlier reporting that the company discussed such features in 2021 and in internal materials this year. Editorial analysis: Companies that prototype identification features for consumer wearables frequently confront elevated scrutiny because these features change the risk calculus for bystanders and amplify regulatory interest. The presence of unshipped code in widely distributed apps increases reputational exposure even when a feature is not active.

What to watch

Observers should track whether Meta publishes a technical or privacy rationale for the code's presence, whether the code reappears in future app versions, and whether regulators or privacy advocates open inquiries following the reporting. Code audits or third-party analyses of subsequent app builds can confirm whether removal was complete or merely temporary. For practitioners, pay attention to how developer tooling, platform permissions, and app store review processes adapt to manage exploratory but sensitive features in consumer-facing apps.

Scoring Rationale

The story matters because unreleased face-recognition code appeared in an app installed on more than 50 million devices, creating a notable privacy and trust incident. It is not a technical breakthrough, but it has significant product governance and regulatory implications for wearable AI.