Android XR smart glasses will blur faces and ban live facial recognition

Faces blurred before Gemini sees them

Google has set out the privacy safeguards built into its Android XR smart glasses ahead of their launch. According to Android Police, citing a post by Android XR vice president and general manager Shahram Izadi, the glasses will automatically blur or redact faces before images are sent to Google's servers for analysis by Gemini, the company's AI assistant.

The protection kicks in whenever the wearer starts a Gemini query, for example asking about their surroundings, translating text on a menu or identifying an object in view. If other people appear in the camera frame, their faces are blurred first. The same rule applies to the wearer: looking in a mirror or pointing the camera at your own face also triggers redaction.

Google adds that images and videos captured for Gemini queries will not be kept in Gemini's history, and that such visuals will not be used to train AI models, personalise adverts or power targeted advertising.

Third-party apps face strict limits

Beyond face blurring, Google has restricted what Play Store apps for the glasses can do. Apps are banned from using the cameras for real-time facial recognition or live streaming. Only first-party apps will support such functions, which lets Google apply its privacy features there.

Developers will also be unable to disable or interfere with the front-facing privacy LED, the small light that tells people nearby when the glasses are recording. Google says it has built in anti-tampering measures: the camera will not record if the indicator is covered or switched off.

Each pair of glasses will also include a dedicated physical switch so wearers can turn the device off completely when it is not in use.

Limits of the safeguards

Android Police notes that Google did not say whether the system is completely foolproof, or whether some images could still reach its servers unblurred. The measures are set against a well-known precedent: a Harvard student modified Meta's Ray-Ban smart glasses into a working real-time facial recognition device, showing how easily such hardware can be misused for surveillance.

Meta's Ray-Ban glasses have faced criticism over privacy, with some members of the public concerned they could be used to film people secretly. The Android XR safeguards appear designed to address those same concerns before launch rather than after.

What the first glasses offer

The first Android XR glasses come from Samsung, produced in partnership with eyewear makers Warby Parker and Gentle Monster. Each pair has a front-facing 12 megapixel ultrawide camera, built-in speakers and a microphone, and up to nine hours of battery life per charge. The charging case allows several top-ups.

The glasses work with Google Pixel and Samsung Galaxy phones running Android 17 or One UI 9, as well as with iPhones. Preorders are open now, with prices starting at $395. Shipping in the United States and South Korea begins on 10 December 2026.

Why the privacy rules matter

Always-on cameras worn on the face raise questions that phone cameras do not, because bystanders often cannot tell when they are being recorded. Hardware signals such as a recording LED, plus a ban on third-party facial recognition, give Google a way to set expectations across the whole ecosystem rather than app by app.

If the measures work as described, they could become a template other makers of smart glasses follow. Readers considering the devices can preorder now, but should note that the details come from Google's own announcements and have not been independently tested.