In a move that feels less like accountability and more like covering tracks, Meta has quietly scrubbed the face recognition components from its Meta AI companion app — one day after WIRED exposed the unreleased system, internally codenamed “NameTag.”

The removal was swift. The version of Meta AI published on the same day as WIRED’s report still contained code libraries explicitly built for face recognition. Friday’s release? Clean. No biometric signature processing. No “Person recognized” alert triggers. No folders for storing cropped images of unidentified faces.

Andy Stone, Meta’s VP of communications, told WIRED the feature is “purely exploratory” and that “no final decision has been made.” But the timeline tells a different story: code was shipping to 50 million+ devices months ago, quietly baked into an app millions already had installed.

What NameTag Was Supposed to Do

According to WIRED’s analysis, NameTag was designed to:

• Capture faces through the smart glasses cameras
• Convert them into unique biometric signatures (faceprints)
• Compare those faceprints against a local database on the user’s device
• Flag matches with a “Person recognized” alert
• Store unrecognized faces — cropped and indexed — for later processing

The implications are staggering. A wearable device that passively identifies people in public, running on hardware already sitting on millions of faces. Privacy advocates immediately flagged the obvious: stalkers, abusers, and surveillance-happy users could weaponize this against strangers with no consent mechanism in sight.

The Gaslighting Was Part of the Plan

The most damning detail? Meta didn’t just ship the code — it publicly gaslit anyone who raised concerns. Andrew Bosworth, Meta’s CTO, called WIRED’s reporting “incredibly misleading” and “absolutely dishonest.” This after Meta refused to answer basic questions: whether it had already built the face profile database, how long biometric data would be retained, whether data would phone home to Meta’s servers, or whether users would get any opt-in/opt-out controls.

And now the code is gone. That raises its own question: was this removal always planned, or did the public scrutiny force a retreat? Meta won’t say.

Leftover Fragments Tell the Story

Not everything was cleaned up. A few remnants remain in the latest build — an internal debug menu label, a dormant profile link pointing to system components that no longer exist. Digital litter from a privacy nightmare that almost was.

Kade Crockford from the ACLU of Massachusetts put it plainly: removal doesn’t undo the decision to ship the code in the first place. They pointed to a Massachusetts consumer privacy bill that passed the state House unanimously last week — legislation that would impose real enforcement, not just voluntary corporate hand-wringing.

The Bigger Picture

This isn’t just a Meta story. It’s a hardware story. Smart glasses are becoming the most intimate surveillance device most people will ever own — a camera that sits on your face, pointed at the world, always on. If the industry wants to avoid a regulatory crackdown, it needs to prove these devices can exist without becoming identification machines.

Meta just proved it can’t be trusted to make that call on its own.

The code may be gone for now. But the infrastructure is there. The hardware is in the wild. And we all know how easy it is to flip a switch back on.

—

Source: WIRED