Meta did not hold a press conference. There was no splashy blog post, no privacy dashboard update asking if you wanted your phone to carry face-scanning software for its smart glasses. The company simply shipped the code. Millions of people downloaded it, tucked inside an app update they never read.
A WIRED investigation published June 4 revealed that Meta buried an unreleased feature called “NameTag” inside the Meta AI companion app, the same software that pairs with its Ray-Ban and Oakley smart glasses. The code has been distributed through app updates since at least January 2026, sitting dormant on millions of phones, converting faces into biometric signatures for on-device matching and wearer alerts. Meta insists the feature is inactive and doesn’t transmit data to its servers. But dormant isn’t absent. Back in February, the New York Times reported that Meta was internally exploring face recognition for its glasses. The difference now is that the groundwork is not locked in a lab. It’s already on your phone, installed while you were waiting for a coffee, updated while you slept, and written into the same terms of service nobody has time to parse.
Why Dormant Code Is Still a Threat
This matters because infrastructure always precedes policy. Once the pipeline exists, the only thing separating a normal Tuesday from a mass ambient surveillance test is a server-side switch. Meta’s glasses already look like ordinary eyewear. Add face recognition, and every coffee shop, subway car, and protest line becomes a potential identity checkpoint. The company will argue that matching happens locally, that biometric data stays on the device, that users will have granular control. Those are engineering details, not ethical guarantees. The social contract changes the moment a stranger can point a camera at your face and see your name without asking. Privacy advocates flagged this exact risk when Meta’s plans surfaced in February. Their fears weren’t speculative; they were prescient. The Independent noted that security researchers are already warning about the hidden surveillance potential baked into the platform. When your hardware is designed to disappear into everyday fashion, the surveillance disappears with it.
Meta knows how this works. Drop the code quietly, let investigators and journalists sound the alarm, then weather the news cycle until the public adjusts. Six months from now, when the feature officially launches, the shock will be gone. The argument will shift from “should this exist?” to “here is how to turn it off.” We’ve seen this movie before. It’s the same playbook that turned always-on microphones and location tracking into background noise. The tech industry loves to frame these moments as inevitable progress. They aren’t. They’re choices, and Meta just chose to build the machinery before asking anyone if they wanted it. While Google and Samsung were busy showing off their own AI glasses at I/O 2026 with at least some public stagecraft, Meta was already seeding the underlying tech onto devices without the theater. That distinction matters. One approach treats the public like adults who deserve a say. The other treats them like endpoints. It’s the same quiet update strategy the company uses across its apps, dropping new tools into Messenger and Instagram before anyone requests them.
What Happens When the Switch Flips
Right now, NameTag is inactive. But the code is present, tested, and waiting. If Meta enables it, even with on-device processing, the implications ripple outward. Protesters, domestic violence survivors, teenagers at a mall; anyone in public view becomes a searchable entry in someone else’s peripheral vision. Regulators will scramble to catch up, but the software is already deployed across millions of devices. The real question is not whether Meta will flip the switch. It’s whether society will still have the energy to push back when the feature is framed as just another helpful AI tool. History suggests we won’t. By the time the toggle appears in settings, buried three menus deep beneath friendly illustrations, the argument will already be lost.
I don’t trust dormant as a comfort. Code that ships is code that runs, and Meta has already proven it will move first and apologize later. Face recognition in smart glasses isn’t a convenience feature; it’s a permanent shift in how we move through public space. It removes the anonymity of a crowd and replaces it with a database you never agreed to join. We should treat it that way before the switch gets flipped, not after. The glasses are already on shelves. The software is already in pockets. All that’s left is the decision to say no, and we’re running out of time to make it.
