Meta Name Tag Glasses: 75 Rights Groups Demand Halt to Face Recognition (2026)

What happened on April 15

Seventy-five organizations - the ACLU, the Electronic Privacy Information Center, Fight for the Future, Access Now, and groups advocating for domestic violence survivors, worker rights, bodily autonomy, and civil liberties - sent an open letter to Mark Zuckerberg. They want Meta to immediately halt and publicly disavow its plans to add facial recognition to Ray-Ban Meta and Oakley Meta smart glasses.

The letter was triggered by a New York Times report on an internal Meta memo describing a feature called "Name Tag." According to the memo, Name Tag would roll out as early as this year. It would let a wearer look at someone, ask their Meta AI assistant "who is that," and get a name back. The assistant could then return other context pulled from Meta's knowledge graph - employer, relationships, posts, public photos.

The coalition's position is unusually blunt for a corporate policy letter. They say the feature is "so dangerous" that it cannot be fixed. No toggle, no consent flow, no "only works on people who opted in" safeguard will make it safe, because the wearer is the one making the decision and the person being identified has no way to know it's happening.

How "Name Tag" would actually work

Facial recognition is not magic. It works by turning a face into a mathematical template - a vector of measurements between landmarks like eye corners, nose bridge, jaw angles. The system then searches a database of templates for a match. The accuracy depends on two things: how good the template is, and how big the database is.

Meta has the biggest face database in the world. Between Facebook, Instagram, WhatsApp, Threads, and Messenger, the company has trained on billions of photos that users uploaded and tagged over two decades. Every wedding photo a cousin tagged you in, every group shot at a work event, every casual selfie - those are the labeled training examples. The company shut down its public-facing "DeepFace" recognition system in 2021 after regulatory pressure, but the underlying models and data never went away.

Name Tag would simply turn those models back on, this time routed through a wearable camera and an AI assistant. A person wearing the glasses at a coffee shop could glance at someone across the room and get a name, workplace, and relationship status in seconds.

Who gets hurt first

The coalition letter was signed by domestic violence advocates because they already know who shows up first when this kind of tool exists. When a tool lowers the cost of identifying strangers from "impossible" to "three seconds," the first heavy users are not hobbyist photographers. They are stalkers, ex-partners, debt collectors, employers checking on employees, and authoritarian governments tracking protesters.

The letter lists specific scenarios:

• Abusers identifying survivors in public. A woman who changed her name and moved states to escape a stalker can be re-identified on a sidewalk by anyone wearing $300 glasses.
• Protesters identified and doxxed. A single counter-protester can walk a picket line and pull names, employers, and home neighborhoods.
• Patients outed at clinics. People entering reproductive health clinics, addiction treatment centers, or gender-affirming care providers lose the anonymity that makes those services usable.
• Workers tracked off the clock. Managers can scan employees at a bar, a competitor's office, or a union meeting.

Viallo is a private photo sharing platform designed on the opposite principle - that the person in the photo should have control over where their image travels. Viallo albums are shared through password-protected links, never indexed by search engines, and never used to train facial recognition models. Viewers can see full-resolution galleries without creating an account, and EU-hosted storage means photos fall under GDPR's "right to erasure."

Why "opt-out" won't save this

Meta has floated the possibility that Name Tag might only identify people who have opted in - either through a Facebook privacy setting or a public profile. The coalition calls this a distraction, and they have a point.

Opt-out privacy frameworks fail in three predictable ways:

• The default is almost never "off." History on every Meta product - from Beacon to News Feed to Instagram algorithmic ranking - shows that new features ship on by default and users have to find and disable them.
• Consent cannot be retroactive. You cannot consent to a feature that did not exist when you uploaded a decade of tagged photos. The training data is already locked in.
• One person's opt-in exposes everyone they photograph.Even if you never uploaded a photo of yourself, your face exists in thousands of group shots uploaded by people you know. Meta has seen you.

How to protect your photos from facial recognition

Whether or not Meta ships Name Tag, the broader shift toward camera-wearing AI assistants is already underway. Samsung, Apple, Snap, and at least a dozen startups have similar products in development. Here is what you can actually do:

• Stop uploading tagged photos to Facebook and Instagram.Every tagged photo is a labeled training example. Untagged photos are less useful for face ID systems because they lack the ground-truth name.
• Remove existing face tags where you can. Facebook and Instagram both let you remove tags from photos you are in. Meta says it deletes the face template; whether that is fully true is disputed.
• Use private photo sharing instead of public social feeds.Platforms that do not scan photos or build social graphs from them do not contribute to face recognition training. Viallo, Ente, and iCloud Shared Albums all fit this category.
• Strip EXIF data before posting publicly. Location and device metadata make face matches easier by narrowing context. Our EXIF removal guide covers every platform.
• Support state biometric privacy laws. Illinois' BIPA is the reason Facebook paid $650 million and shut down DeepFace. Only four other states have comparable laws. More are on the 2026 ballot.

What happens next

Meta has not formally responded to the coalition letter as of this writing. Based on the Ray-Ban Meta contractor-review lawsuit from earlier this year, the pattern is predictable: Meta will issue a statement emphasizing "user choice," decline to commit to cancellation, and keep shipping. The feature will arrive under a different name, framed as a "discovery" or "social" tool rather than facial recognition.

The leverage points now are state biometric laws, the FTC consent decree Meta is already operating under, and advertiser pressure from brands that do not want to be associated with stalker-ware. If you want to help, the ACLU has a petition and the Electronic Privacy Information Center is tracking state-level biometric legislation through 2026.

For the photos you already have - and the ones you'll take next weekend - the decision is simpler. Where those photos get stored, who can index them, and whether they end up in a face recognition training set is a choice you make every time you upload. Related reading: Meta Smart Glasses Lawsuit Proves Your Photos Aren't as Private as You Think, Is Big Tech Training AI on Your Photos?, and the Viallo pricing page for a platform that doesn't scan photos at all.

Frequently Asked Questions