A facial recognition system hiding in plain sight
Meta’s NameTag system was a dormant facial recognition feature embedded in a smart glasses companion app update that converted faces into numerical faceprints and matched them to stored databases without users or bystanders being clearly informed or meaningfully able to consent.
The core issue is not that Meta experimented with facial recognition, but that Meta facial recognition code shipped to more than 50 million phones before anyone outside the company knew it existed. Buried in the app that supports the Ray-Ban line, NameTag used three AI models in sequence: one to detect faces, one to generate a 2,048-number faceprint for each face, and one to compare those faceprints to a stored database in real time. EFF researchers were able to activate this dormant NameTag system in debug mode and watch it correctly identify a known face, prompting them to call it “nearly ready to go” and warn it could turn the public into a “distributed surveillance network.” The lesson is blunt: once such code is quietly on your phone, the switch from dormant to live is trivial—and we only heard about it because someone went looking.
Consent theater in a world of invisible cameras
NameTag was not a hypothetical lab experiment; it was deployed code that could identify people from photos or live views without their consent or knowledge. In a market where Meta’s Ray-Ban smart glasses already account for more than seven million units and roughly 82 percent of global smart glasses sales, the impact of such a feature is obvious. The wearer gets effortless recognition; everyone else becomes a scanned object. The system’s design—face detection, faceprint creation, database matching—was built for frictionless identification, not for meaningful, informed opt-in.
Real-world behavior already shows how fragile smart glasses privacy is. A peer-reviewed study surveying more than a thousand owners found that 17 percent had recorded others without consent or in prohibited locations. When nearly one in five users are already pushing boundaries, secretly shipping a near-ready facial recognition pipeline amplifies the harm. According to that study’s authors, biometric data is treated as sensitive information under existing law, yet regulators admit that practical consent mechanisms for smart glasses in public spaces are “unworkable at scale.” In other words, the legal system has no realistic answer for being passively scanned every time you walk down a street.
Meta’s pattern: apologize with silence, innovate with spectacle
When the Electronic Frontier Foundation published its analysis on June 4, Meta responded not with a public explanation, but with code removal. On June 5, an update stripped the NameTag system from the app without a press release or detailed communication to the people who had unknowingly hosted it. Meta has not said it will never bring the feature back; it removed the code, and those are not the same promise. This quiet rollback sits atop a stack of earlier concerns: earlier demonstrations showed that Meta Ray-Ban glasses could already be paired with external tools to identify strangers and retrieve names, home addresses, jobs and even fragments of Social Security numbers from public databases.
Meanwhile, Meta’s hardware strategy is pushing smart glasses deeper into everyday life. The company’s new in-house glasses line starts at USD 299 (approx. RM1,400), the lowest entry price yet for AI-enabled, camera-equipped eyewear, while another major player recently unveiled Specs AR glasses priced at USD 2,195 (approx. RM10,100). Meta sold more than seven million Ray-Ban units in 2025 alone, more than tripling year-over-year. At the same time, Meta is loudly rolling out transparency tools on its platforms and algorithmic controls on Instagram, but that transparency does not extend to experimental, privacy-sensitive code shipped to people’s phones. The message is mixed: the feed is negotiable; the camera on your face is not.
New Instagram tools, old surveillance risks
Adding to the dissonance, Meta is actively promoting new Instagram Stories creation tools tailored for Ray-Ban Meta, Oakley Meta and other Meta glasses. Spin View lets viewers rotate their phones to explore a wider field of view from the wearer’s perspective, Multi-Cam merges simultaneous footage from phone and glasses, and new editing tools allow reframing, background noise reduction, and speed control for glasses-captured clips. Media from these devices is flagged with a subtle glasses icon, and users can open an overview and point-of-view shooting guide to refine their content. This is smart product design: the tools make the devices more fun, more useful, and more immersive.
But the timing is telling. On one track, Meta is normalizing always-on cameras through colorful Stories formats that showcase how seamlessly your life can be streamed from your face. On the other, it quietly ships and then removes Meta facial recognition code discovered only because researchers inspected the app. The experience for everyone not wearing the glasses is the same: more recording, more sharing, more potential for misuse. When 17 percent of owners already admit to recording without consent, enhancing creation tools without equally visible, enforceable safeguards looks less like innovation and more like marketing gloss over a widening surveillance gap.
What happens when the next NameTag ships?
The NameTag saga highlights a deeper, unresolved problem: there is no clear line between experimental features and deployed capabilities in Meta’s hardware ecosystem. Today, NameTag is gone; tomorrow, NameTag-style code could reappear in another update after the news cycle moves on, when the glasses cost USD 299 (approx. RM1,400) and sit on fourteen million faces instead of two million. Other tech giants are confirming their own smart glasses for the coming months, signaling that face-worn cameras with AI overlays are becoming a mainstream category, not a niche gadget.
The industry has learned the wrong lesson from earlier, more conspicuous devices: make the hardware invisible and the backlash will fade. Yet the central question remains unanswered by both companies and regulators: what concrete commitments will prevent another covert facial recognition system from slipping into an app used by tens of millions of people? Until there are binding rules on feature disclosure, explicit bans on unconsented public identification, and independent auditing of privacy-sensitive updates, smart glasses privacy will depend on luck and watchdogs. Meta’s NameTag incident is not an isolated misstep; it is a preview of the surveillance infrastructure we sleepwalk into if we treat invisible cameras as harmless fashion accessories.
