Exclusive: Meta Secretly Built Facial Recognition for Smart Glasses Into Its Massively Downloaded AI App, WIRED Analysis Finds

A deep technical analysis of Meta’s software conducted by WIRED has revealed the company has stealthily integrated facial recognition technology built for its smart glasses line into a mobile application already downloaded by millions of users globally.

Code added discreetly to the Meta AI app across multiple 2024 updates confirms the tool, internally codenamed “NameTag,” identifies people captured by the smart glasses’ on-board camera, and alerts the wearer when it matches a captured face to a known entry once activated.

The discovery of NameTag hidden in the public, live build of the Meta AI app confirms Meta began shipping core facial recognition code to end-users’ phones while publicly framing the technology as something the company was still “thinking through.” Back in April 2024, Meta stated that if it ever deployed facial recognition, it would not launch the tool without first taking “a very thoughtful approach.” But WIRED’s investigation found core components of the NameTag system were already integrated into software distributed to millions of people as early as January 2024.

While the feature is not yet enabled for end users, NameTag lives inside the Meta AI companion app — which has been downloaded more than 50 million times, and is required to access key features of Meta’s smart glasses, including its Ray-Ban and Oakley branded models. If fully activated, NameTag will convert faces captured by Meta’s glasses into unique biometric signatures (commonly called faceprints), then cross-reference each new capture against faceprints stored locally on the user’s phone. This local database is currently configured to receive ongoing updates from Meta. Recognized faces trigger a push notification to the wearer, while unmatched faces are automatically cropped, indexed, and saved to a folder marked “pending.”

| Got a confidential tip? |

|---|

| Are you a current or former Meta employee with insight into the company’s facial recognition work? WIRED wants to hear from you. Contact the reporter securely via Signal at dmehro.89 or dell.3030 using a personal, non-work device. |

NameTag revives a category of technology Meta publicly announced it was shutting down back in 2021, when the company said it would delete more than 1 billion faceprints belonging to Facebook users after years of controversy over its automatic photo-tagging system. Meta ultimately paid a $650 million settlement to resolve a class-action lawsuit brought by Illinois users over unauthorized biometric collection, and in 2024 agreed to a separate $1.4 billion settlement with the state of Texas over allegations of unlawful biometric data harvesting.

Meta’s renewed push for consumer-facing facial recognition comes amid growing opposition to the technology, with privacy advocates warning it puts a dangerous surveillance tool directly in the hands of bad actors ranging from stalkers to federal immigration agents. Internal Meta documents published by The New York Times in February 2024 showed the company planned to roll out the feature during a “dynamic political environment,” when Meta believed its biggest critics would be distracted by other issues.

WIRED’s analysis confirms three AI models that power NameTag have already been deployed from Meta’s servers and now reside on customers’ phones, a finding that was independently reproduced by outside technical experts. One model detects faces in camera footage, one crops the detected faces to isolate them, and a third encodes the face into usable biometric data.

Only partial traces of the user interface currently exist in the app code, offering clues about how the feature will ultimately work for consumers. A May 2024 version of the app rebrands the feature for end users as “Connections,” and invites users to “remember the people you met.” It remains unclear whose faces will be included in the system’s recognition database, how user profiles are created, or how many people could ultimately be identifiable through the tool.

WIRED shared its findings with two outside security researchers who separately examined the Meta AI app and reproduced key parts of the outlet’s analysis: Cooper Quintin, a security researcher and senior public interest technologist with the nonprofit Electronic Frontier Foundation’s Threat Lab, and independent security and privacy researcher Buchodi (who publishes under a pseudonym) and has more than a decade of experience reverse engineering consumer software and surveillance technologies.

“The feature is not yet exposed to consumers but seems nearly ready to go,” Quintin said. “Despite the billions of reasons not to, Meta seems to have created the capacity to turn their customers into a distributed surveillance machine.”

Buchodi ran additional independent tests on the full facial recognition processing pipeline. (Their full technical analysis is available here.) To test whether the matching system worked as intended, Buchodi added a single faceprint (taken from an image of deceased French philosopher Michel Foucault) to the app’s local gallery. After triggering NameTag with a separate image of Foucault, the app generated a notification reading: “Person recognized.”

“The main components of a face-recognition feature are already in Meta's companion app,” Buchodi said. “Not many pieces stand between this and a working feature.”

In April 2024, more than 70 advocacy groups — including the American Civil Liberties Union, Electronic Privacy Information Center, and Fight for the Future — called on Meta to scrap the NameTag project entirely, warning it would let stalkers and abusers silently identify strangers in public spaces. “Our competitors offer this type of face-recognition product, we do not,” a Meta spokesperson said in a statement to WIRED at the time. “If we were to release such a feature, we would take a very thoughtful approach before rolling anything out.”

Privacy advocates argue that embedding facial recognition into a mass-market wearable platform would normalize a capability Meta previously pulled back from amid widespread privacy concerns.

“You're setting norms and standards by putting technology into the ecosystem,” said Joseph Jerome, a former Meta Reality Labs policy official who led privacy reviews for the company’s AR and VR products, commenting on Meta’s outsized influence on the wearable tech industry. “I don't know how Meta can responsibly deploy a technology like this.”

“Regardless of any sensational reporting, the facts are simple: We've said before we're exploring these types of features, and what you're seeing is just evidence of that exploration,” said Meta spokesperson Ryan Daniels. “Nothing has shipped to consumers and no final decision has been made on what to do here, if anything. If we do decide to roll something out, we will take a thoughtful approach and do so with full transparency. One decision we can be clear about—we are not building a central face database.”

WIRED’s code review confirms the NameTag system is currently designed to pull faceprints from Meta’s servers and store them locally on individual user devices.

Meta’s first large-scale consumer facial recognition system, launched by Facebook in 2010, analyzed uploaded photos and suggested tags for people appearing in users’ images. It quickly scaled to more than 1 billion users, becoming one of the largest consumer facial recognition systems ever deployed.

The technology drew scrutiny almost immediately. European regulators and U.S. privacy advocates questioned its legality as early as 2011, with ongoing concerns over whether users had given meaningful consent for the creation of their biometric data. In 2019, Meta paid a $5 billion settlement to the Federal Trade Commission and Department of Justice to resolve a broad privacy case that included claims related to unregulated facial recognition.

In November 2021, Meta announced it would shut down the original system and delete all face templates it had built, citing growing societal concern over the role of facial recognition. But that public decision was never viewed internally as a permanent retreat from the technology, said Jerome, who joined Reality Labs in mid-2021. “There was always this tension of, well, when do we roll back out face recognition?”

According to internal documents reviewed by The New York Times, Meta planned to debut facial recognition on its smart glasses in 2025, first rolling it out to attendees of a conference for people with visual impairments before launching it to the general public. The launch never happened. But the technology addresses a proven real-world demand: Existing assistive devices already let blind users identify faces they have personally enrolled, and a 2018 study of blind users by Cornell Tech and Facebook researchers found every participant ranked facial recognition as an important daily task.

Meta did not respond to questions about which users might be identifiable through NameTag; whether it intends for photos, faceprints, or other data generated by the system to ever be transmitted back to Meta’s servers; or whether the company plans to let users opt in to the feature rather than opting out. EssilorLuxottica, which manufactures Meta’s Ray-Ban and Oakley smart glasses, did not respond to a request for comment.

Woodrow Hartzog, a privacy law professor at Boston University, says even opt-in protection — should Meta eventually offer it — is thin. Consent, he notes, is often tied to access to a job, a benefit, or core service features. Framing privacy as a matter of personal choice benefits businesses, he says, placing no meaningful limits on data collection while letting companies claim users are in full control.

“We know that the more these systems are deployed, the more people come to see them as unexceptional,” Hartzog said. “And the more we come to see them as unexceptional and routine, the more people tend to start to take their moral cues about whether it's desirable or good to have your face scanned. That's just human psychology.”