Meta Strips Facial Recognition From Smart Glasses App

A pair of smart glasses on a wooden surface, the kind of device that just lost its built-in face recognition feature, for now — Photo via Unsplash

TL;DR: On June 4, 2026, WIRED reported that Meta had quietly shipped facial-recognition code to millions of smartphones inside the Meta AI app that pairs with Ray-Ban and Oakley smart glasses. The internal "Name Tag" feature converts every face in the glasses' field of view into a 2,048-number faceprint array, then matches it against a database the user can populate. EFF Threat Lab had reverse-engineered the same code on June 4 and published the proof. By June 5, Meta's app update had quietly removed the code. The Electronic Frontier Foundation declared "VICTORY" on June 8. PetaPixel, Engadget, Malwarebytes, and Gizmodo all covered the reversal. EFF's tracker lists the win. Read past the headline, though. The code that processed faces on millions of phones for months can be redeployed whenever Meta decides the time is right. The privacy LED on the glasses is software-controlled and can be disabled. The Kenya data-labeling scandal is still unresolved. The April coalition letter is still unanswered. The Texas Attorney General is still investigating. Public pressure worked. The architecture is still broken.

What Happened

On June 4, 2026, WIRED published a story that should not have been a surprise to anyone following the file. Reporter Dhruv Mehrotra, in collaboration with two outside security researchers, ran static code analysis on the Meta AI app shipped to millions of Ray-Ban Meta and Oakley Meta glasses owners. The findings: facial-recognition code, internally called "Name Tag," was embedded in the app, functional, and just waiting for a server-side switch flip.^[1]

Name Tag is the same feature EFF Threat Lab security researcher Cooper Quintin had reverse-engineered the same day in a piece titled "Move Fast, Surveil Things." Quintin activated the feature in debug mode and watched the glasses identify a known face in real time. Meta's spokesperson at the time called the code "evidence of exploration, not intent." Exploration does not ship to millions of phones. Meta's "exploration" processed faces on those phones for at least several months before WIRED's report.^[2][3]

The publication triggered the response EFF and the 75-organization civil society coalition had been building since March. Within 24 hours, Meta had pushed a new build of the Meta AI app to the App Store and Google Play that quietly removed the facial-recognition code. The company told outlets the change was voluntary, a privacy-driven decision made before WIRED's report. The timing is the story.^[1][2][4][5]

On June 8, 2026, the EFF published a "VICTORY" deeplink under the headline "Meta Strips Facial Recognition Code From Smart Glasses App After Public Outcry." The April 13 ACLU-led 75-organization letter was credited in the framing. WIRED published a follow-up on the same day. PetaPixel, Malwarebytes, and Gizmodo all picked up the story in the days that followed. EFF's "Move Fast, Surveil Things" piece was the proof of concept. The April letter was the political cover. WIRED's report was the trigger. Meta's PR team read the room.^{[4][5][6][7][8]}

Why This Is a Half-Win at Best

EFF called it a victory. The EFF is correct. The activist machinery of consumer-rights groups, the privacy bar, the civil-liberties nonprofits, and the reporters willing to do static analysis on shipped code worked. Meta, a $1.4-trillion company with the most well-funded privacy-violation legal team in the history of advertising technology, pulled a feature before it shipped. That is not nothing.

It is also not structural. Four reasons.

1. The code is still on the phones. Most users will not update the Meta AI app. Most users do not know there is a Meta AI app separate from the glasses. The code that was in the app yesterday is still in many of those apps today, dormant in the build, waiting to be re-enabled by a single server-side flag. The infrastructure for face recognition on millions of phones is now a meta-distributed system. Meta can re-enable it whenever it wants.^[1][2]

2. The LED privacy indicator can be disabled. The recording LED on the Ray-Ban Meta glasses is a hardware switch tied to the camera. Disabling it requires physical disassembly. The facial-recognition LED, the one that would tell a bystander the glasses are running Name Tag in front of their face, is software-controlled. Research published earlier in 2026 demonstrated that a sufficiently motivated user can disable it with a software flag. The bystander has no signal that the glasses are running face recognition. They have no signal that the glasses are recording, either. The hardware is not built for the threat model that the EFF, the senators, the Texas AG, and the April coalition have been warning about.^[3][9]

3. The Kenya data-labeling scandal is still unresolved. The April 1, 2026 investigation by Sweden's SVT and subsequent coverage by TechWeez documented Meta contractors in Nairobi reviewing Ray-Ban Meta footage that included users in bathrooms, bedrooms, and other intimate settings. Kenya's Office of the Data Protection Commissioner opened a formal investigation on April 1, 2026. As of June 12, 2026, no public findings, no fine, no public remediation plan. The same human-review pipeline that produced those privacy violations is still in place. The only thing that changed this week is that the feature feeding the pipeline with a new class of biometric data (faces) is paused, not the pipeline itself.^[10][11]

4. The April coalition letter is still unanswered. The 75 organizations, led by the ACLU, the ACLU of Massachusetts, and the New York Civil Liberties Union, sent a joint letter to Meta, Congress, the FTC, and the DOJ on April 13, 2026 demanding the company abandon its plan to add facial recognition to smart glasses. The letter was not answered. Senators Markey, Wyden, and Merkley sent a separate letter on March 17 with an April 6 deadline. The deadline passed. Meta did not answer. The 75 organizations, the senators, the Texas AG, the Illinois BIPA plaintiffs, the BIPA class action: none of them have stopped. The face-recognition fight is paused, not over.^{[12][13][14][15]}

The Architecture That Put the Code on Millions of Phones Is Still in Place

To understand why a pulled code release is not a fix, walk through the chain of decisions that put the code there in the first place.

Meta's product roadmap, leaked internally in February 2026, said the company planned to add facial recognition to smart glasses and "release the feature quietly to avoid public scrutiny." TechCrunch reported the memo the same day. The 75-organization ACLU-led letter that followed on April 13, joined by EFF, EPIC, Public Citizen, Consumer Reports, and dozens of others, was a direct response to the memo.^[16][17]

The senators' letter, the April 6 deadline, the Texas AG investigation launched in May, the Illinois BIPA class action, the RSAC live demo where a researcher identified strangers using facial recognition on Meta glasses in front of an audience, the EFF reverse-engineering, the WIRED report: each of these is a pressure event. Each one is also a proof point. The roadmap Meta said in February, the one that promised to ship facial recognition "quietly to avoid public scrutiny," is the same roadmap that produced the silent code release in May, the same one that produced the WIRED report in June, and the same one that produced the Meta pull-out this week.^[18][19][20]

Pulling the code is the end of this round. It is not the end of the fight. The next round starts the next time Meta decides the press cycle is friendly enough to flip the switch.

How the Pressure Machine Actually Worked

WIRED's June 4 report is the trigger. It is not the cause. The cause is the year of pressure that made the WIRED report land.

February 13, 2026: TechCrunch reports on Meta's internal plan to add facial recognition to Ray-Ban Meta glasses, citing a leaked product memo that said the feature should be released "quietly to avoid public scrutiny."^[16]

February 13, 2026: The EFF publishes "Seven Billion Reasons Facebook Should Abandon Its Face Recognition Plans," a reference to the $1.4 billion Meta paid to Texas in a 2024 biometric-privacy settlement. The deeplink frames the smart-glasses rollout as the second coming of a program Meta publicly abandoned in 2021.^[18]

March 17, 2026: Senators Markey, Wyden, and Merkley send a letter to Meta CEO Mark Zuckerberg demanding transparency on the facial-recognition rollout. They give Meta until April 6 to answer six direct questions, including whether users will be able to upload photos of friends and family to build personal faceprint databases, and whether Meta will share biometric data with law enforcement.^[13]

March 31, 2026: A coalition of civil-society organizations, organized by the Consumer Federation of America and UltraViolet Action, sends a letter to Congress, the FTC, the DOJ, Meta, and EssilorLuxottica (Meta's glasses partner) demanding Meta abandon the rollout. The coalition includes EPIC, Public Citizen, Consumer Reports, the American Federation of Teachers, and others. They call the feature a "creepy and unacceptable escalation of surveillance."^[18]

April 1, 2026: Sweden's SVT publishes its investigation into Meta's Kenya data-labeling operation. Kenya's Office of the Data Protection Commissioner opens a formal investigation. TechWeez and other outlets pick up the story.^[10][11]

April 13, 2026: The ACLU, joined by the ACLU of Massachusetts, the New York Civil Liberties Union, and 72 other organizations (75 total), sends a follow-up letter to Meta, expanding the demand to include a halt to all biometric data collection on smart glasses, not just facial recognition. Kade Crockford of the ACLU of Massachusetts calls the rollout "a red line society must not cross."^[12]

May 2026: Texas Attorney General Ken Paxton opens an investigation into Meta's smart glasses, citing potential violations of Texas's Capture or Use of Biometric Identifier Act (CUBI). The investigation follows the same legal theory as the 2022 settlement that produced the $1.4 billion Meta paid to Texas over Facebook face-tagging.^[14]

June 4, 2026: The EFF publishes "Move Fast, Surveil Things," Cooper Quintin's static-analysis report. The same day, WIRED publishes "Meta Silently Added Face-Recognition Code for Its Smart Glasses to Millions of Phones" by Dhruv Mehrotra, independently confirming Quintin's findings with the help of two outside researchers. The Verge, Engadget, and the tech press pick it up overnight.^[1][2][3]

June 5, 2026: Meta's app update quietly removes the facial-recognition code from the Meta AI app. The change is not announced. PetaPixel and Malwarebytes notice on June 9. Gizmodo follows on June 11.^[6][7][8]

June 8, 2026: WIRED publishes a follow-up titled "Meta Deletes Face-Recognition System From Its Smart Glasses App After WIRED Report." The EFF publishes a "VICTORY" deeplink the same day, crediting the 75-organization ACLU-led letter and EFF's own static-analysis work. The EFF tracker entry for the smart-glasses facial-recognition fight is updated to "WIN."^[4][5][9]

What Stays Broken

Walk through EFF's own tracker. The "smart glasses facial recognition" item is the only line that flipped this week. The other items on the tracker did not move.^[9]

The Kenya investigation is open. The Texas AG investigation is open. The Illinois BIPA class action (filed in March 2026) is in discovery. The 75-organization letter to Congress is unanswered. The senators' letter to Meta is unanswered past its April 6 deadline. The Markey press conference in March, where he called the rollout "a stalker's dream," is on the record. The 75-organization coalition is still active. EFF's own "Move Fast, Surveil Things" piece is still the public proof that the code processed faces on millions of phones for months before Meta said a word.^{[2][9][12][13][14][15]}

Pulling the code is a Meta decision, not a Meta capability loss. The code can be re-pushed. The feature can be re-enabled. The roadmap can be re-launched. The next launch attempt will likely be quieter, more legal-cover, more "privacy safeguards" framing, more "user opt-in" language. The reason it will likely be quieter is exactly because the round that just ended worked.

What You Can Do Today

Update the Meta AI app anyway. Even if you do not own Ray-Ban Meta glasses, the app is on your phone. Pull the latest build. The facial-recognition code is removed in the new build, but other surveillance and ad-targeting code is not. The fewer apps with faceprint capability on your device, the better.
Do not upload face photos to Meta services. The faceprint database is user-populated. Every photo you upload to Facebook, Instagram, Messenger, or WhatsApp is, in principle, training data for a future faceprint matching pipeline. The 2021 settlement that "ended" Facebook's face-tagging program was not actually an end. The same data and the same matching infrastructure are still there.
Support the EFF and the coalition orgs. The Consumer Federation of America, UltraViolet Action, EPIC, ACLU, Public Citizen, Consumer Reports, EFF: these are the groups that did the year-long organizing that made the WIRED report land. They are donor-funded. They need money. The next round of the fight will be a state-by-state biometric-privacy push and a federal trade-practices-rule campaign at the FTC. That work is not free.
Watch the Texas AG investigation. The 2022 Texas settlement that produced the $1.4 billion Meta payout was the most expensive biometric-privacy enforcement action in U.S. history. Paxton's 2026 investigation is the second shoe. The 2022 settlement required Meta to stop face-tagging without consent. The 2026 investigation will likely require Meta to stop face-recognition in smart glasses without a warrant. Watch the docket.
Push for an Illinois-style biometric privacy law in your state. Illinois BIPA is the only state law strong enough to make a class action against Meta economically rational. Texas CUBI is the only state law that has produced a billion-dollar Meta settlement. The ACLU-led 75-organization coalition is lobbying state legislatures. Find your state, find your legislator, push for a bill.
Treat any smart glasses around you as a recording device. The hardware LED on the Ray-Ban Meta glasses can be disabled. The recording LED is software-controlled on the facial-recognition side. The glasses can be recording, face-recognizing, and uploading for AI training without the person next to you knowing. The legal regime around bystander recording is a patchwork. The safest assumption is that any face in front of a smart-glasses wearer is now data.
Read the EFF tracker. EFF maintains a running tracker of the smart-glasses facial-recognition fight. The "WIN" line that flipped this week is one line. The other lines are still open. The next round will land on that tracker before it lands in your news feed.^[9]

The Bottom Line

EFF won a round. The 75-organization coalition won a round. WIRED did the reporting that made the round close. The April ACLU letter, the senators, the Texas AG, the Kenya investigation, the EFF static analysis: the pressure machine worked.

The code is pulled. The launch is paused. The political cost of flipping the switch in 2026 is now too high. Meta will come back, with a different framing, a different launch surface, a different legal team, and the same face-recognition infrastructure. The next round will be a state biometric-privacy fight, a federal FTC trade-rule fight, and a BIPA class-action discovery fight. The code on your phone is pulled. The architecture that put it there is the same architecture that will put it back.

The privacy LED can still be disabled. The Kenya investigation is still open. The April letter is still unanswered. The Texas AG is still investigating. The 75-organization coalition is still organized. The fight is paused, not won. The work is not over. The surveillance is not over. The camera in the glasses is not off. It is just not on, for now.