Overview
- Journalists at WIRED found unreleased face‑recognition code called NameTag inside the Meta AI companion app that would convert faces into biometric “faceprints” and match them against a database on a user’s phone.
- Less than 48 hours after WIRED’s disclosure, Meta pushed an app update that independent analysts including the EFF verified removed most of the face‑recognition libraries and storage folders.
- Meta’s executives called the reporting misleading and said the work was exploratory and not shipped to consumers, but the company declined to answer detailed questions about whether any face profiles or retained test data exist.
- Researchers reported small fragments of NameTag remained in the updated app and that the original code included logic to crop, index, and locally store images of unrecognized faces for future processing.
- The episode strengthens scrutiny from regulators and advocates given Meta’s recent biometric settlements and raises practical risks for bystanders, women’s safety, stalking, and state surveillance if always‑on ID ever becomes operational.