Google's Smart Glasses Hides Your Face From Gemini, But Not From a Simple Recording

Google says its new smart glasses redact bystanders' faces when Gemini is asked a question, but that protection vanishes the moment someone just hits record.
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.

If you're standing near someone wearing Google's new smart glasses, here's the uncomfortable truth: whether your face gets protected depends entirely on what the wearer is doing in that moment, and you have no way to tell the difference.[1]

Google opened preorders for its new Android XR smart glasses on November 12, with sales beginning December 10.

Developed with Samsung, Warby Parker and Gentle Monster, the glasses arrive amid growing scrutiny of Meta's competing eyewear, including complaints about covert recording and reports of facial recognition development. Google says its own privacy protections extend to people "on both sides of the lens."[2][3]

When someone uses the glasses to ask Gemini about their surroundings, such as translating a menu, Google says nearby faces are automatically redacted before the AI processes the image.

But the person doing the recording certainly still has your face.

In this article
When Google's face-blurring protection doesn't apply
What Google is promising, and what it isn't
The rules wearers have to agree to
What you can actually do about it

When Google's face-blurring protection doesn't apply

Here's the catch, and it's significant for anyone standing in frame: face redaction only applies when the wearer is actively querying Gemini with the camera. It doesn't apply to ordinary photos or videos the wearer takes and saves to their own gallery, which CNN confirmed directly with Google. Point the glasses at someone and just hit the camera button, and there's no blurring, no redaction, nothing. Your face gets captured exactly the way it would on any camera.

A bystander standing on a sidewalk has no way to know which mode is active. A recording light illuminates when the wearer takes a photo or video for the gallery, and Google says sensors are designed to detect if that light is covered or tampered with, blocking capture if so.

"It's not perfect," said Juston Payne, Google's senior director of product management for XR, about the tamper-detection system.

Even a perfectly functioning indicator light tells you almost nothing useful. Calli Schroeder, director of the AI and human rights program at the Electronic Privacy Information Center, told CNN that a person might not feel comfortable asking a stranger to stop recording, or a worker might not want to confront their boss over it.

Consequently, bystanders have no actual mechanism for consent, light or no light.

Janusz Swierczynski, who researches human interaction with emerging tech at Oxford's Saïd Business School, added that these protections still mostly address privacy for the individual wearer and not the people who end up in the frame, partly because the technology is moving faster than policy can keep up.

Visually impaired bystanders also have unique difficulties here, as some can't see the light at all.

What Google is promising, and what it isn't

It's worth being precise about what Google has actually committed to versus what remains a stated design goal. The company says data stored on the glasses is encrypted, and data sent to a phone or the cloud is encrypted in transit.

Ars Technica reported that the data isn't encrypted while sitting on Google's own servers.

Google also says it plans to bring "Private AI Compute," a more secure processing environment, to the glasses in the future. That hasn't happened yet.

None of these features have been independently tested or verified by outside researchers. They're Google's own account of how the system is supposed to work, not findings from a security audit.

Meta's glasses have faced their own scrutiny, including a security researcher's demonstration of how the glasses could be hacked, and Norway has weighed a temporary ban on facial-recognition-capable smart glasses over similar concerns.

Google also says the microphones on its glasses are designed to focus on the wearer's voice by default rather than picking up nearby conversations, using a "denoiser step" to filter out ambient speech, according to WIRED. Recording for photos and video is capped at 90 seconds.

The rules wearers have to agree to

Google also requires users to agree not to record people without permission, film in sensitive places or cover the recording light. Violating those rules could result in the glasses being disabled. Users must be at least 18 years old to set up the device.

Google Play policies also prohibit apps from using the glasses' camera for real-time facial identification or social livestreaming. Apps can't interfere with the recording light, and Google Ads and Shopping prohibit promoting accessories designed to conceal or disable it.

Izadi told CNN that Google has no plans to add facial recognition to its glasses now or in the future, a direct contrast with Meta's reported work on facial recognition for its own eyewear, which a New York Times report surfaced earlier this year and which heightened privacy concerns across the category.

What you can actually do about it

If you use Gemini glasses yourself, you have some real controls. You can grant or revoke camera and microphone access for individual apps at any time. In the Gemini Apps Privacy Hub, you can review and delete individual chats, wipe your entire Gemini history, and control whether your chat text gets used to improve AI models. And every pair of glasses has a physical power switch that fully shuts off the camera and microphones, letting you wear them as ordinary glasses.

If someone else is wearing the glasses, watch for the recording light. It can tell you when a photo or video is being saved, although it won't illuminate during ordinary Gemini visual queries. You can ask someone to stop recording, but Google's privacy settings don't let bystanders prevent their faces from appearing in saved footage. Efforts to establish stronger legal protections have also faced setbacks, including California's vetoed smart glasses legislation.

Author Details
Thomas Kent is a multi-disciplined reporter with over a decade of experience covering online platforms, digital trends, and consumer-facing tech. Tom focuses on digital privacy, data tracking, and user behavior, with a particular interest in how cookies, online surveillance, and platform design shape the modern internet experience. His reporting takes a research-driven, news-focused approach, translating complex technical topics into clear, accessible insights.

Citations

[1] Designing for privacy on intelligent eyewear

[2] Here's How Google Says It's Making Its Smart Glasses More Private

[3] Google knows smart glasses have a privacy problem. Here's how it's trying to fix it