Meta Quietly Built — Then Deleted — Face-Recognition Feature in Smart Glasses

Meta removed face-recognition code from its Ray-Ban Meta smart glasses on or around June 8, 2026, after WIRED found the unreleased feature hidden in the companion app. Engadget confirmed the removal the same day. The code was never turned on for users and was not listed as something the glasses could do — but its presence in the app was enough to make Meta remove it once the story went public.

What Was Found

The face-recognition code was sitting dormant, or switched off, inside Meta's AI Glasses companion application. People who review software code as part of their job — a standard practice in technology reporting — looked at the app's inner workings and found strings of code and references that pointed to a face-recognition capability the company had not announced.

The Ray-Ban Meta glasses already use your camera to recognize things you're looking at and provide real-time information and reminders. Think of it this way: the glasses have a built-in pipeline — a pathway — from the camera through to the internet and back. Adding face recognition on top of that existing pipeline would be technically straightforward, like adding a new layer to a structure that already exists.

Meta's public privacy page for the glasses does not mention face recognition at all. The page talks about protections that remove identifying details when you use the camera features. The gap between what Meta said publicly and what engineers found in the code is what made this story important.

What Meta Did About It

Meta removed the code without explaining why it had been there in the first place or what it was supposed to do. That kind of quiet removal, rather than a public explanation, is worth paying attention to.

Here is why: Meta operates under close watch from government regulators because of its history with data privacy issues. When a company finds unreleased features in its code that the public does not know about, there is a question about whether the company has a responsibility to tell anyone. In most places, the answer today is no — as long as the feature was never actually turned on. But once the code becomes public and the company deletes it instead of explaining it, that answer becomes harder to defend.

The Ray-Ban Meta glasses do have a physical button to turn them on and off. Meta says this button is an important safety feature for privacy. Whether a physical on-off switch is enough protection against a camera that could identify people's faces is something regulators in the United States and Europe have been thinking about for years without reaching a clear answer.

Why This Matters More Than You Might Think

Face recognition on your smartphone requires you to open an app and point the camera at something. Face recognition built into glasses that you wear all day — glasses that other people might not even realize are cameras — is a completely different problem. The person wearing the glasses can identify anyone silently, without them knowing or being able to stop it. That is the core concern.

This is not just a theoretical worry. In late 2024, students at Harvard built a working proof-of-concept using Ray-Ban Meta glasses, a livestream, and publicly available face-recognition software to identify strangers in real time. They assembled these pieces without needing anything inside Meta's actual glasses — they used tools already available to anyone. The fact that Meta had face-recognition code inside its own app suggests the company was at least exploring building this capability natively.

We have seen this pattern before. When Google Glass launched in 2013, Google explicitly blocked face-recognition apps from working on the device after public pressure. At the time, many observers saw this as a business decision to keep the product acceptable to society, not as a permanent technical or legal rule. Google Glass never became popular with the public anyway. But now smart glasses are back, this time millions of people are using them, and the face-recognition question has returned — and we still do not have clear rules about what is allowed.

What Meta Says vs. What the Code Suggests

Meta's privacy page describes protections designed to strip away identifying information before it gets stored. If those protections work as described — and if outside experts were allowed to check them — that would be meaningful protection. But we cannot verify either of those things from outside the company. The dormant face-recognition code does not prove Meta was actively using these protections to hide a secret program, but it does raise questions about whether we can trust the privacy assurances Meta makes.

For corporate IT departments and security teams, this incident is a practical problem. Ray-Ban Meta glasses are showing up in offices where employees use them for work. But unlike phones, which companies can monitor and control with management software, there is no good way for a company to audit what the glasses' camera or app are doing. That is a structural gap in the technology, and incidents like this one highlight it for buyers who had not thought it through before.

What Comes Next

Meta has not explained where the face-recognition code came from, how long engineers had been working on it, or whether it will show up again in a future version with proper user permission. Those three questions matter most to privacy researchers, regulators, and developers building tools for Meta's glasses.

The EU AI Act, which started enforcement in 2025, prohibits real-time facial identification in public spaces except in very limited cases. Whether hidden code in a consumer app counts as something companies have to report to regulators is exactly the kind of gray area regulators and lawyers are still working through. The United States has no equivalent national rule, though some states like Illinois have their own biometric privacy laws that might apply depending on where data is processed.

There is a more hopeful way to look at this. The system worked at a basic level: independent reporting found an undisclosed capability, the company removed it under public pressure, and the incident will probably push both regulators and Meta to tighten their own internal rules for features that use facial recognition or similar biometric data. That is a form of accountability, even if it only happened after the fact rather than by design.

But there is still a harder question that the industry has not answered since Google Glass: what rules should govern face-recognition features on wearable devices before companies ship them to the public — not after journalists find the code.