Face Recognition Should Not Ship Quietly
WIRED and EFF found unreleased face-recognition code in Meta's smart-glasses platform. The privacy problem is not only what shipped. It is what can be switched on later.
A phone app can hide a lot of future in plain sight.
That is the part that bothers me about the Meta smart glasses story. WIRED reported that Meta silently added unreleased face-recognition code for its smart-glasses platform to millions of phones. The system, according to WIRED, is designed to identify people through biometric data stored on the user's phone.
EFF's Threat Lab said it confirmed that the code is present through static analysis.
Meta may call it unreleased. Maybe it never turns on in this form. Maybe it gets renamed, narrowed, consent-wrapped, region-locked, policy-reviewed, and buried under five layers of product language.
Fine. But dormant face recognition in a consumer glasses platform is still a privacy event. You do not need the feature to be live before people are allowed to object to the direction of travel.
Glasses change the social contract
A phone camera is already awkward enough. At least you usually know when someone is holding up a phone.
Glasses are different. They sit on the face. They point where the person looks. They turn the act of walking through a room into something that can feel like passive capture. Add face recognition and the device stops being just a camera. It becomes a name tag reader for the world.
That is not a small product feature.
It changes what it means to be in public, in a shop, at a protest, on a date, in a clinic waiting room, or just standing next to someone who likes new gadgets too much.
The usual defense is that people can consent. That sounds tidy until you imagine the actual scene. Someone wearing smart glasses walks into a cafe. Everyone else is now supposed to understand the model, firmware, app permissions, cloud settings, face-recognition state, data retention policy, and whether the wearer changed anything yesterday.
Nobody is doing that. Not in real life.
Local data does not make it harmless
WIRED's description says the system is designed around biometric data stored on the user's phone. That matters. Local processing can be better than shipping every face to a server.
But "local" is not a magic privacy word.
If the device can identify people, the risk already exists at the edge. It can change how the wearer behaves. It can leak through logs, screenshots, contact matching, notifications, assistant summaries, or future integrations. It can also normalize the idea that the person wearing the computer gets to know more about everyone else than everyone else agreed to reveal.
That is the asymmetry. The glasses owner opts in. The bystanders inherit the consequences.
Europe should be especially allergic to this. Biometric identification is not just another app permission under a cute icon. It is sensitive data tied to your body. Once face recognition becomes ambient, the harm is hard to undo because you cannot rotate your face like a password.
The timing is the warning
The code showed up before the feature was public.
That is normal software development, yes. Apps contain experiments, flags, prototypes, and half-built features all the time. I am not pretending every dormant code path is a scandal.
But face recognition deserves a higher bar. If a company with Meta's history is putting that capability into a glasses ecosystem, the public should hear about it from the company before researchers and reporters have to dig it out of the app.
This is where transparency actually matters. Not a polished announcement after the decision is already made. Not a privacy whitepaper written like a legal escape room. A plain answer: what can the device identify, when, with whose consent, where is the biometric data stored, how is it deleted, and what stops the feature being expanded later?
If those answers are not ready, the feature is not ready either.
What to do if you use these devices
If you own Meta smart glasses or the companion app, do the boring privacy audit:
- Keep the app updated, but review permissions after major updates.
- Disable camera, microphone, contacts, location, and notification access unless you actually need them.
- Check whether the app has access to photos, people, or contact-matching features you forgot about.
- Avoid wearing camera glasses in places where people cannot meaningfully opt out.
- If you manage devices for a workplace, treat smart glasses as recording equipment, not as ordinary eyewear.
For everyone else, the answer is more political than technical. Regulators should not wait until ambient face recognition is everywhere before asking basic questions. Companies should not get to seed biometric capability into consumer devices quietly and then frame criticism as fear of innovation.
I am not against smart glasses. I am against pretending that a camera on your face is just another screen.
Face recognition makes the device about everyone in the room, not only the person who bought it. That means everyone in the room deserves a say.