Meta Glasses Are About To Remember Your Face, Not Just Record It

Meta’s new patent could let AI-powered glasses decide what’s worth recording, identify people, and build a detailed picture of your social life.
We receive compensation from the products and services mentioned in this story, but the opinions are the author's own. Compensation may impact where offers appear. We have not included all available products or offers. Learn more about how we make money and our editorial policies.

Imagine a pair of smart glasses with a camera that uses AI to decide what’s worth recording. Laugh too hard or surprise someone wearing a pair of Meta glasses, and your face could end up in Meta’s database. No need for the owner's consent, let alone the consent of the people being recorded.

That's the future Meta has just quietly filed a patent for. Meta glasses could soon get a new feature that records people or moments and then creates a highlights package, even without the user specifically instructing it to do so.[1]

Worse still, the feature could also use facial recognition to remember people, build a social profile, and gather extremely private information, such as who the people close to you are and who are mere acquaintances.

These glasses have already been dubbed "pervert glasses" and touted as a privacy nightmare. They’ve, in fact, also been banned in several popular pubs, restaurants, and theaters in the UK.

Here’s a deep dive into Meta’s new patent, the privacy concerns around it, and what you can do to protect yourself from unwarranted and non-consensual monitoring.

In this article
What does Meta’s new patent entail
Why Meta glasses’ new feature is a privacy nightmare
Meta could turn your relationships into data
What can you do to protect yourself
Bottom line

What does Meta’s new patent entail

Meta’s patent, published on August 13, discusses an AI assistant that watches through your camera, figures out what’s interesting, and automatically captures the moment for you, later creating a highlight for your viewing.

How and when it decides what’s worth capturing is easily the biggest Black Mirror-esque thing about it.

  • The system uses machine-learning models to decide what might be interesting — the owner doesn’t necessarily have to instruct the system what to capture.
  • It can use signals like facial recognition to decide who’s in the frame. For instance, it could differentiate between someone close to you, such as your wife, and other acquaintances.
  • Additionally, it can detect facial expressions. For example, people who are smiling or laughing might be considered an interesting moment to capture.
  • Similarly, it could also detect objects. For instance, a convertible car might be considered more interesting than a parking garage.
  • Once the system finds its point of interest, it can zoom or crop the camera view so that the subject is centered while the background and other objects are blurred out.
  • After this, the AI assistant will filter the content based on quality and make a personalized selection based on user preferences, profile data, and relationship data.

The patent cites an example where the AI assistant asks, “I have generated some highlights of tonight’s dinner party. Would you like to see them?”

In doing so, the system also includes a memory recall feature. For instance, you could ask questions like, “Did I meet this guy in a white T-shirt before?” and it will scan for episodic memories to find relevant past media.

The patent also includes a feature called “ambiguity disambiguation.” This means that whenever a user gives an ambiguous voice command such as, “Take a picture of that,” the device can use camera data to figure out what that refers to and capture or save the media after showing the user a cropped preview.

Why Meta glasses’ new feature is a privacy nightmare

There are three major privacy concerns with Meta’s proposed new feature.

No bystander consent

As has been the case with Meta’s glasses, bystander consent seems to be missing from this patent. The AI picks up “visual cues” such as happy or smiling faces before capturing a moment.

However, you could be completely unrelated to the person wearing the glasses and still end up as a subject of the recordings, all without your explicit consent.

This is the same privacy concern due to which several pubs, restaurants, and theaters in the UK have banned Meta’s smart glasses, stating that these glasses impose “surreptitious surveillance” which no one should be subjected to.

Though the patent states there will be an opt-in option for the feature, this can only be on the user’s side and not for those who are being recorded by these glasses.

An AI that’s always looking for something

The second and perhaps the biggest issue is its proactive nature — the camera is always looking for visual cues to record.

This means it’s silently active in the background even when it’s not allegedly recording anything. There’s little clarity on whether this always-on feature stores any data.

How long Meta will remember you for

Lastly, there’s the inherent problem with its memory recall feature.

Because the AI could recall previous memories and answer questions like, “Have I seen this guy before?”, it raises the question of how long Meta stores such sensitive data. Again, this has not been laid out in the patent.

An investigation by Svenska Dagbladet, a Swedish newspaper, found that a Kenyan-based subcontractor was reviewing footage from Meta’s smart glasses, which included sensitive content such as people using the toilet, engaging in sexual activity, and nude images, all without the users’ consent. Meta is now facing a lawsuit for this.

Meta's own policy states that photos and videos stored in the cloud are kept for 30 days, though Meta says it does not use this content to train its AI. Separately, voice transcripts and recordings can be retained for up to a year to help improve Meta's AI products.

Not long back, Meta attempted to add a similar facial recognition feature to the Meta AI app, its companion app for Meta smart glasses. According to Wired, it could convert facial images into unique biometric signatures to identify strangers in public settings and could later trigger person-recognized alerts, similar to the memory recall feature in the current patent.

However, after huge public backlash, Meta had to quietly discontinue this feature with an update on June 5. At that time, the Electronic Frontier Foundation warned that this deletion doesn't mean a permanent change of heart. And the current patent potentially goes on to prove this claim.

Meta could turn your relationships into data

What's concerning is that this new patent goes way beyond just facial recognition and builds a social graph. For instance, it could distinguish your spouse from mere acquaintances and use relationship data to prioritize highlights.

This means that it’s not just recognizing faces, but also building a database of people you are close to and how much time you spend with them. This is much richer and more sensitive data than your personally identifiable information.

In the wrong hands, this data could be used to build an incredibly detailed picture of your personal life, making you far more vulnerable to targeted phishing, social engineering, and other forms of manipulation.

Meta had earlier emphasized that its glasses come with an LED light that blinks whenever recording is on. If anyone tries to block or modify the LED, the recording stops working altogether.

However, as per a report from the Cognitive Privacy Project, many people either miss the small light or assume that it is a Bluetooth indicator. It’s clearly not enough.

Combined with AI security cameras and even your car collecting personal data about you, we’re already at the gates of a dystopian, around-the-clock surveillance state.

What can you do to protect yourself

Meta has only filed a patent as of now, which means it could be a while before we actually see smart glasses with this feature.

When it does take to the streets, though, there's little you could do about it, as it’s the manufacturer’s feature.

Even if it comes with opt-in functionality, it entirely depends on the user if they enable or disable it. You, as a bystander, would have little say in what those glasses are recording.

  1. But if you're using these glasses, you can work towards improving everyone’s privacy around you by opting out of this automatic recording functionality.
  2. Furthermore, make sure that you take proper consent before recording anyone. Or, if you notice someone recording you without consent, you could politely ask them not to do so.
  3. To stay safe from phishing attacks, you must never click on suspicious links in emails or SMS messages, no matter how urgent or tempting they sound.
  4. It’s also a good idea to get an identity theft protection service that will continuously scan online databases, including the dark web, and alert you if it finds your personally identifiable information.

Bottom line

Meta seems to be paying no attention to all the criticism its “pervert glasses” have been attracting. Instead, it has gone ahead and filed a new patent that takes surveillance to a new level, throwing consent completely out of the window.

The new proposed feature could record moments it considers interesting, even without the user's consent or prompt, and later convert these captured moments into highlights.

And the company’s proposed privacy protection — an opt-in option — is, unfortunately, merely a gimmick. It still leaves a lot of room for discretion on the part of the user.

Make sure that you obtain explicit consent from the people around you when donning these glasses. Or, if you find anyone recording you, you could politely ask them to delete any such captured files. To further avoid your data’s misuse, get an identity theft protection service.

4.8
Editorial Rating
Get Deal
On DeleteMe's website
2026 Editors’ Choice
Best Data Removal for Businesses
Privacy Protection
DeleteMe
PROMOTION: Use the Code PARTNER20 for 20% Off
  • Data removal service that covers 89–986 sites and re-scans every quarter to catch anything that reappears
  • Sends quarterly privacy reports showing what info was found, which brokers had your data, and how long each removal took
  • Includes email masking so you can share a stand-in address instead of your real one

Author Details
Krishi Chowdhary specializes in digital privacy, cybersecurity, and consumer technology. He has written extensively on online privacy tools and broader cybersecurity topics, including online scams, data breaches, age verification, and emerging digital threats. Krishi believes technology reporting should empower readers, not confuse them, and is committed to making even the most technical subjects easy to understand without compromising on accuracy or depth. His work has appeared in leading technology publications, including CNET, ExpressVPN, and TechRadar, where he has covered topics ranging from cybersecurity incidents and privacy product announcements to artificial intelligence and major technology news

Citations

[1] Meta’s patent for smart cameras enabled by assistant systems