A coalition of more than 25 civil liberties organizations has sent a pointed letter to Meta Platforms Inc., demanding the company take concrete steps to prevent its Ray-Ban smart glasses from becoming tools for real-time facial recognition surveillance. The warning, led by the American Civil Liberties Union and joined by groups including the Electronic Frontier Foundation, Amnesty International, and Fight for the Future, lands at a moment when the technical barriers separating a consumer gadget from a mass surveillance device have essentially collapsed.
The letter isn’t hypothetical hand-wringing. It’s a response to something that already happened.
Last fall, two Harvard students — AnhPhu Nguyen and Caine Ardayfio — built a system called I-LAMA that paired Meta’s Ray-Ban smart glasses with the facial recognition search engine PimEyes and large language models to identify strangers in real time. Walk past someone on the street, and within seconds the system could pull up their name, home address, phone number, and Social Security number. The students demonstrated the tool on unsuspecting people in public, then published their findings to raise awareness. They did not release the code. But the proof of concept was devastating in its simplicity — no specialized hardware, no government database access, just commercially available products stitched together with a few lines of code, as Mashable reported.
That demonstration crystallized fears that had been simmering since Meta first released its camera-equipped Ray-Ban Stories in 2021. The latest generation of the glasses, launched in partnership with EssilorLuxottica in late 2023, added Meta’s AI assistant and livestreaming capabilities. They look almost indistinguishable from regular Ray-Bans. That’s the problem. Unlike a phone camera, which requires a visible gesture to operate, the glasses can record and stream video with no outward indication to the people being captured. A tiny LED light is supposed to signal when recording is active, but it’s small enough to miss in daylight and trivial to obscure.
The coalition’s letter, dated June 2025, asks Meta to do several things. Disable the glasses’ ability to livestream video to third-party facial recognition services. Implement technical safeguards that would prevent the camera feed from being piped into identification systems. Publicly commit to never integrating facial recognition into the glasses’ native software. And engage with civil rights organizations on governance frameworks before shipping future hardware updates.
Meta has so far responded with what amounts to a policy deflection. A company spokesperson told multiple outlets that Meta’s terms of service already prohibit using its products for facial recognition surveillance and that the company is “committed to building AI responsibly.” The coalition finds this insufficient. Terms of service are enforced after the fact, if at all. They don’t prevent a determined user — or a government agency — from repurposing the hardware.
“The issue is not what Meta’s policies say. The issue is what Meta’s technology enables,” the ACLU wrote in a summary accompanying the letter. The organization has long argued that privacy protections built on policy rather than architecture are functionally meaningless when the underlying technology makes abuse trivially easy.
And the technology is getting easier by the month. PimEyes, the facial recognition engine used in the Harvard demonstration, charges as little as $29.99 per month for individual searches against a database of billions of scraped images. Clearview AI, the controversial startup that built its database by scraping social media platforms including Facebook, has contracts with more than 3,100 law enforcement agencies in the United States. The raw infrastructure for identifying anyone, anywhere, already exists. What Meta’s glasses add is the capture layer — a socially invisible, always-available camera sitting on someone’s face.
This convergence has attracted regulatory attention beyond the advocacy community. In the European Union, the AI Act that took partial effect in 2025 includes strict prohibitions on real-time biometric identification in public spaces, with narrow exceptions for law enforcement. Several U.S. cities, including San Francisco, Portland, and Boston, have enacted local bans on government use of facial recognition. But no federal law in the United States restricts private individuals from using facial recognition technology on other people in public. That gap is precisely what the coalition is asking Meta to address through product design rather than waiting for legislation that may never come.
The timing of the letter also reflects growing unease about Meta’s broader AI ambitions for wearable hardware. CEO Mark Zuckerberg has repeatedly described smart glasses as the successor to the smartphone — a computing platform people will wear all day, every day. Meta’s roadmap includes adding a display to future versions of the Ray-Ban glasses, effectively creating augmented reality spectacles that overlay digital information on the physical world. Internal prototypes, reported on by The Verge, show a product called Orion that can project holograms, translate languages in real time, and recognize objects. Adding face identification to that stack would be technically trivial and, from a product standpoint, almost irresistible.
Consider the user experience Meta could offer. You’re at a conference. You can’t remember the name of the person walking toward you. Your glasses identify them, pull up their LinkedIn profile, remind you that you last spoke at a dinner in 2023, and surface their recent posts. Useful? Enormously. Dangerous? Also enormously. The same capability deployed by a stalker, an authoritarian government, or an abusive ex-partner produces outcomes that are categorically different from a networking convenience.
The ACLU’s letter specifically flags the risk to protesters, undocumented immigrants, and domestic violence survivors — populations for whom anonymity in public spaces is not a luxury but a safety requirement. During the 2020 racial justice protests, law enforcement agencies used facial recognition to identify demonstrators from social media footage. Smart glasses would make that process orders of magnitude more efficient, removing the need to review footage after the fact and enabling identification at the moment of encounter.
Meta isn’t the only company building camera-equipped wearables. Snap’s Spectacles, now in their fifth generation, include dual cameras and AR displays. Google is developing smart glasses under its Project Astra initiative. Apple’s Vision Pro, while a bulky headset today, is widely expected to shrink toward a glasses form factor over the coming decade. But Meta is furthest along in consumer adoption. The company said in late 2024 that Ray-Ban Meta glasses were its fastest-selling hardware product ever, with sales surpassing expectations in every quarter since launch. That installed base is what makes the facial recognition question urgent now rather than theoretical.
So what could Meta actually do? The coalition suggests several technical approaches. One is on-device content analysis that detects when the camera feed is being routed to a facial recognition API and blocks the connection. Another is cryptographic watermarking of the video stream that would allow downstream services to identify and reject footage from Meta glasses. A third is simply limiting the livestreaming capability to approved platforms that have their own anti-facial-recognition policies. None of these are foolproof. A sufficiently motivated adversary can always find workarounds. But the coalition’s argument is that raising the technical barrier matters, even if it doesn’t eliminate the risk entirely.
There’s a precedent for this kind of design-level intervention. When Apple introduced AirTags in 2021, stalking concerns prompted the company to build in anti-tracking features — alerts when an unknown AirTag is traveling with you, automatic sound emissions after separation from the owner’s device. Those measures weren’t perfect. Stalkers found ways around them. But Apple’s willingness to constrain its own product’s functionality in response to misuse set a standard that other companies have since followed. The coalition is essentially asking Meta to do the same thing: accept that a product’s potential for harm can outweigh its commercial utility and engineer accordingly.
Meta’s incentives point in the other direction. The company has invested billions in its Reality Labs division, which develops the glasses, VR headsets, and AR prototypes. Reality Labs has lost more than $50 billion cumulatively since 2020, a figure that Zuckerberg has defended as a long-term bet on the next computing platform. Restricting the glasses’ capabilities — particularly features that would make them more useful and more likely to be worn all day — cuts against the strategic logic of that investment. Every constraint is a reason for a consumer to leave the glasses in a drawer.
But the reputational and regulatory risks of inaction may be larger. Meta already faces ongoing litigation and regulatory scrutiny over its data practices. The Federal Trade Commission’s 2024 proposed order against the company included restrictions on how it can use minors’ data. The EU’s Digital Markets Act imposes interoperability and data-sharing requirements. Adding a facial recognition scandal to that list — particularly one involving physical-world surveillance rather than online data collection — could trigger legislative responses that constrain not just Meta but the entire wearable computing industry.
The Harvard students who built I-LAMA understood this dynamic. In their writeup, they explicitly framed the project as a warning rather than a product. “We built this to show what’s possible today, with tools anyone can access, so that the conversation about regulation happens before the technology is everywhere,” Nguyen said in an interview published by Mashable. That conversation is now happening. Whether it produces meaningful change depends largely on whether Meta treats the coalition’s letter as a PR problem to manage or a design challenge to solve.
The distinction matters more than it might seem. PR responses — blog posts about responsible AI, vague commitments to “ongoing dialogue” with stakeholders — cost nothing and change nothing. Design responses require engineering resources, product trade-offs, and executive willingness to say no to features that users might want. Meta has historically been better at the former than the latter. Its track record on privacy, from the Cambridge Analytica scandal to the repeated FTC consent decrees, suggests a company that treats user protection as a compliance exercise rather than a product principle.
But the glasses are different from a social media platform in one critical respect. A Facebook privacy violation involves data you chose to put online. A facial recognition violation via smart glasses involves data extracted from your physical presence without your knowledge or consent. You didn’t sign up. You didn’t agree to terms of service. You were just walking down the street. That asymmetry — between the person wearing the glasses and the person being seen by them — is what makes the civil liberties argument so forceful. And it’s what makes Meta’s policy-only response so inadequate.
The next few months will be telling. Meta’s annual Connect conference, typically held in the fall, is expected to showcase new hardware and AI features for the glasses line. If the company announces expanded camera capabilities, always-on AI processing, or deeper integration with third-party services without corresponding privacy safeguards, the coalition has signaled it will escalate — pushing for legislative action at the state and federal level. Several state legislatures, including those in Illinois, Massachusetts, and Washington, already have biometric privacy laws that could be amended to address wearable devices specifically.
For now, the letter sits on Meta’s desk. Twenty-five organizations. A clear set of demands. A technical demonstration proving the threat is real. What happens next will say a great deal about whether the wearable computing industry can self-correct before regulators force corrections that may be far blunter and far more restrictive than anything the ACLU is currently asking for.
The glasses are already on people’s faces. The question is whose face gets identified next.
WebProNews is an iEntry Publication